TRANSFORM_MATRIX Use the android.colorCorrection.transform matrix and android.colorCorrection.gains to do color conversion. All advanced white balance adjustments (not specified by our white balance pipeline) must be disabled. If AWB is enabled with `android.control.awbMode != OFF`, then TRANSFORM_MATRIX is ignored. The camera device will override this value to either FAST or HIGH_QUALITY. FAST Color correction processing must not slow down capture rate relative to sensor raw output. Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied. If AWB is enabled with `android.control.awbMode != OFF`, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run). HIGH_QUALITY Color correction processing operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate) Advanced white balance adjustments above and beyond the specified white balance pipeline may be applied. If AWB is enabled with `android.control.awbMode != OFF`, then the camera device uses the last frame's AWB values (or defaults if AWB has never been run). The mode control selects how the image data is converted from the sensor's native color into linear sRGB color.

When auto-white balance (AWB) is enabled with android.control.awbMode, this control is overridden by the AWB routine. When AWB is disabled, the application controls how the color mapping is performed. We define the expected processing pipeline below. For consistency across devices, this is always the case with TRANSFORM_MATRIX. When either FULL or HIGH_QUALITY is used, the camera device may do additional processing but android.colorCorrection.gains and android.colorCorrection.transform will still be provided by the camera device (in the results) and be roughly correct. Switching to TRANSFORM_MATRIX and using the data provided from FAST or HIGH_QUALITY will yield a picture with the same white point as what was produced by the camera device in the earlier frame. The expected processing pipeline is as follows: ![White balance processing pipeline](android.colorCorrection.mode/processing_pipeline.png) The white balance is encoded by two values, a 4-channel white-balance gain vector (applied in the Bayer domain), and a 3x3 color transform matrix (applied after demosaic). The 4-channel white-balance gains are defined as: android.colorCorrection.gains = [ R G_even G_odd B ] where `G_even` is the gain for green pixels on even rows of the output, and `G_odd` is the gain for green pixels on the odd rows. These may be identical for a given camera device implementation; if the camera device does not support a separate gain for even/odd green channels, it will use the `G_even` value, and write `G_odd` equal to `G_even` in the output result metadata. The matrices for color transforms are defined as a 9-entry vector: android.colorCorrection.transform = [ I0 I1 I2 I3 I4 I5 I6 I7 I8 ] which define a transform from input sensor colors, `P_in = [ r g b ]`, to output linear sRGB, `P_out = [ r' g' b' ]`, with colors as follows: r' = I0r + I1g + I2b g' = I3r + I4g + I5b b' = I6r + I7g + I8b Both the input and output value ranges must match. Overflow/underflow values are clipped to fit within the range.

HAL must support both FAST and HIGH_QUALITY if color correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY should generate the same output. 3 3 A color transform matrix to use to transform from sensor RGB color space to output linear sRGB color space. Unitless scale factors

This matrix is either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX. In the latter case, the camera device may round the matrix to account for precision issues; the final rounded matrix should be reported back in this matrix result metadata. The transform should keep the magnitude of the output color values within `[0, 1.0]` (assuming input color values is within the normalized range `[0, 1.0]`), or clipping may occur. The valid range of each matrix element varies on different devices, but values within [-1.5, 3.0] are guaranteed not to be clipped.

4 Gains applying to Bayer raw color channels for white-balance. Unitless gain factors

These per-channel gains are either set by the camera device when the request android.colorCorrection.mode is not TRANSFORM_MATRIX, or directly by the application in the request when the android.colorCorrection.mode is TRANSFORM_MATRIX. The gains in the result metadata are the gains actually applied by the camera device to the current frame. The valid range of gains varies on different devices, but gains between [1.0, 3.0] are guaranteed not to be clipped. Even if a given device allows gains below 1.0, this is usually not recommended because this can create color artifacts.

The 4-channel white-balance gains are defined in the order of `[R G_even G_odd B]`, where `G_even` is the gain for green pixels on even rows of the output, and `G_odd` is the gain for green pixels on the odd rows. If a HAL does not support a separate gain for even/odd green channels, it must use the `G_even` value, and write `G_odd` equal to `G_even` in the output result metadata. OFF No aberration correction is applied. FAST Aberration correction will not slow down capture rate relative to sensor raw output. HIGH_QUALITY Aberration correction operates at improved quality but the capture rate might be reduced (relative to sensor raw output rate) Mode of operation for the chromatic aberration correction algorithm. android.colorCorrection.availableAberrationModes

Chromatic (color) aberration is caused by the fact that different wavelengths of light can not focus on the same point after exiting from the lens. This metadata defines the high level control of chromatic aberration correction algorithm, which aims to minimize the chromatic artifacts that may occur along the object boundaries in an image. FAST/HIGH_QUALITY both mean that camera device determined aberration correction will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality aberration correction algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying aberration correction. LEGACY devices will always be in FAST mode.

n List of aberration correction modes for android.colorCorrection.aberrationMode that are supported by this camera device. Any value listed in android.colorCorrection.aberrationMode

This key lists the valid modes for android.colorCorrection.aberrationMode. If no aberration correction modes are available for a device, this list will solely include OFF mode. All camera devices will support either OFF or FAST mode. Camera devices that support the MANUAL_POST_PROCESSING capability will always list OFF mode. This includes all FULL level devices. LEGACY devices will always only support FAST mode.

HAL must support both FAST and HIGH_QUALITY if chromatic aberration control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

OFF The camera device will not adjust exposure duration to avoid banding problems. 50HZ The camera device will adjust exposure duration to avoid banding problems with 50Hz illumination sources. 60HZ The camera device will adjust exposure duration to avoid banding problems with 60Hz illumination sources. AUTO The camera device will automatically adapt its antibanding routine to the current illumination condition. This is the default mode if AUTO is available on given camera device. The desired setting for the camera device's auto-exposure algorithm's antibanding compensation. android.control.aeAvailableAntibandingModes

Some kinds of lighting fixtures, such as some fluorescent lights, flicker at the rate of the power supply frequency (60Hz or 50Hz, depending on country). While this is typically not noticeable to a person, it can be visible to a camera device. If a camera sets its exposure time to the wrong value, the flicker may become visible in the viewfinder as flicker or in a final captured image, as a set of variable-brightness bands across the image. Therefore, the auto-exposure routines of camera devices include antibanding routines that ensure that the chosen exposure value will not cause such banding. The choice of exposure time depends on the rate of flicker, which the camera device can detect automatically, or the expected rate can be selected by the application using this control. A given camera device may not support all of the possible options for the antibanding mode. The android.control.aeAvailableAntibandingModes key contains the available modes for a given camera device. AUTO mode is the default if it is available on given camera device. When AUTO mode is not available, the default will be either 50HZ or 60HZ, and both 50HZ and 60HZ will be available. If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then this setting has no effect, and the application must ensure it selects exposure times that do not cause banding issues. The android.statistics.sceneFlicker key can assist the application in this.

For all capture request templates, this field must be set to AUTO if AUTO mode is available. If AUTO is not available, the default must be either 50HZ or 60HZ, and both 50HZ and 60HZ must be available. If manual exposure control is enabled (by setting android.control.aeMode or android.control.mode to OFF), then the exposure values provided by the application must not be adjusted for antibanding. Adjustment to auto-exposure (AE) target image brightness. Compensation steps android.control.aeCompensationRange

The adjustment is measured as a count of steps, with the step size defined by android.control.aeCompensationStep and the allowed range by android.control.aeCompensationRange. For example, if the exposure value (EV) step is 0.333, '6' will mean an exposure compensation of +2 EV; -3 will mean an exposure compensation of -1 EV. One EV represents a doubling of image brightness. Note that this control will only be effective if android.control.aeMode `!=` OFF. This control will take effect even when android.control.aeLock `== true`. In the event of exposure compensation value being changed, camera device may take several frames to reach the newly requested exposure target. During that time, android.control.aeState field will be in the SEARCHING state. Once the new exposure target is reached, android.control.aeState will change from SEARCHING to either CONVERGED, LOCKED (if AE lock is enabled), or FLASH_REQUIRED (if the scene is too dark for still capture).

OFF Auto-exposure lock is disabled; the AE algorithm is free to update its parameters. ON Auto-exposure lock is enabled; the AE algorithm must not update the exposure and sensitivity parameters while the lock is active. android.control.aeExposureCompensation setting changes will still take effect while auto-exposure is locked. Some rare LEGACY devices may not support this, in which case the value will always be overridden to OFF. Whether auto-exposure (AE) is currently locked to its latest calculated values.

When set to `true` (ON), the AE algorithm is locked to its latest parameters, and will not change exposure settings until the lock is set to `false` (OFF). Note that even when AE is locked, the flash may be fired if the android.control.aeMode is ON_AUTO_FLASH / ON_ALWAYS_FLASH / ON_AUTO_FLASH_REDEYE. When android.control.aeExposureCompensation is changed, even if the AE lock is ON, the camera device will still adjust its exposure value. If AE precapture is triggered (see android.control.aePrecaptureTrigger) when AE is already locked, the camera device will not change the exposure time (android.sensor.exposureTime) and sensitivity (android.sensor.sensitivity) parameters. The flash may be fired if the android.control.aeMode is ON_AUTO_FLASH/ON_AUTO_FLASH_REDEYE and the scene is too dark. If the android.control.aeMode is ON_ALWAYS_FLASH, the scene may become overexposed. Similarly, AE precapture trigger CANCEL has no effect when AE is already locked. When an AE precapture sequence is triggered, AE unlock will not be able to unlock the AE if AE is locked by the camera device internally during precapture metering sequence In other words, submitting requests with AE unlock has no effect for an ongoing precapture metering sequence. Otherwise, the precapture metering sequence will never succeed in a sequence of preview requests where AE lock is always set to `false`. Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AE updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended: 1. Starting in auto-AE mode: 2. Lock AE 3. Wait for the first result to be output that has the AE locked 4. Copy exposure settings from that result into a request, set the request to manual AE 5. Submit the capture request, proceed to run manual AE as desired. See android.control.aeState for AE lock related state transition details.

OFF The camera device's autoexposure routine is disabled. The application-selected android.sensor.exposureTime, android.sensor.sensitivity and android.sensor.frameDuration are used by the camera device, along with android.flash.* fields, if there's a flash unit for this camera device. Note that auto-white balance (AWB) and auto-focus (AF) behavior is device dependent when AE is in OFF mode. To have consistent behavior across different devices, it is recommended to either set AWB and AF to OFF mode or lock AWB and AF before setting AE to OFF. See android.control.awbMode, android.control.afMode, android.control.awbLock, and android.control.afTrigger for more details. LEGACY devices do not support the OFF mode and will override attempts to use this value to ON. ON The camera device's autoexposure routine is active, with no flash control. The application's values for android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration are ignored. The application has control over the various android.flash.* fields. ON_AUTO_FLASH Like ON, except that the camera device also controls the camera's flash unit, firing it in low-light conditions. The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and may be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE ON_ALWAYS_FLASH Like ON, except that the camera device also controls the camera's flash unit, always firing it for still captures. The flash may be fired during a precapture sequence (triggered by android.control.aePrecaptureTrigger) and will always be fired for captures for which the android.control.captureIntent field is set to STILL_CAPTURE ON_AUTO_FLASH_REDEYE Like ON_AUTO_FLASH, but with automatic red eye reduction. If deemed necessary by the camera device, a red eye reduction flash will fire during the precapture sequence. The desired mode for the camera device's auto-exposure routine. android.control.aeAvailableModes

This control is only effective if android.control.mode is AUTO. When set to any of the ON modes, the camera device's auto-exposure routine is enabled, overriding the application's selected exposure time, sensor sensitivity, and frame duration (android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration). If one of the FLASH modes is selected, the camera device's flash unit controls are also overridden. The FLASH modes are only available if the camera device has a flash unit (android.flash.info.available is `true`). If flash TORCH mode is desired, this field must be set to ON or OFF, and android.flash.mode set to TORCH. When set to any of the ON modes, the values chosen by the camera device auto-exposure routine for the overridden fields for a given capture will be available in its CaptureResult.

5 area_count List of metering areas to use for auto-exposure adjustment. Pixel coordinates within android.sensor.info.activeArraySize Coordinates must be between `[(0,0), (width, height))` of android.sensor.info.activeArraySize

Not available if android.control.maxRegionsAe is 0. Otherwise will always be present. The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAe. The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array. The weight must be within `[0, 1000]`, and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region. The weights are relative to weights of other exposure metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored. If all regions have 0 weight, then no specific metering area needs to be used by the camera device. If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

Only constrains auto-exposure (AE) algorithm, not manual control of android.sensor.exposureTime and android.sensor.frameDuration.

IDLE The trigger is idle. START The precapture metering sequence will be started by the camera device. The exact effect of the precapture trigger depends on the current AE mode and state. CANCEL The camera device will cancel any currently active or completed precapture metering sequence, the auto-exposure routine will return to its initial state. Whether the camera device will trigger a precapture metering sequence when it processes this request.

This entry is normally set to IDLE, or is not included at all in the request settings. When included and set to START, the camera device will trigger the auto-exposure (AE) precapture metering sequence. When set to CANCEL, the camera device will cancel any active precapture metering trigger, and return to its initial AE state. If a precapture metering sequence is already completed, and the camera device has implicitly locked the AE for subsequent still capture, the CANCEL trigger will unlock the AE and return to its initial AE state. The precapture sequence should be triggered before starting a high-quality still capture for final metering decisions to be made, and for firing pre-capture flash pulses to estimate scene brightness and required final capture flash power, when the flash is enabled. Normally, this entry should be set to START for only a single request, and the application should wait until the sequence completes before starting a new one. When a precapture metering sequence is finished, the camera device may lock the auto-exposure routine internally to be able to accurately expose the subsequent still capture image (`android.control.captureIntent == STILL_CAPTURE`). For this case, the AE may not resume normal scan if no subsequent still capture is submitted. To ensure that the AE routine restarts normal scan, the application should submit a request with `android.control.aeLock == true`, followed by a request with `android.control.aeLock == false`, if the application decides not to submit a still capture request after the precapture sequence completes. Alternatively, for API level 23 or newer devices, the CANCEL can be used to unlock the camera device internally locked AE if the application doesn't submit a still capture request after the AE precapture trigger. Note that, the CANCEL was added in API level 23, and must not be used in devices that have earlier API levels. The exact effect of auto-exposure (AE) precapture trigger depends on the current AE mode and state; see android.control.aeState for AE precapture state transition details. On LEGACY-level devices, the precapture trigger is not supported; capturing a high-resolution JPEG image will automatically trigger a precapture sequence before the high-resolution capture, including potentially firing a pre-capture flash.

OFF The auto-focus routine does not control the lens; android.lens.focusDistance is controlled by the application. AUTO Basic automatic focus mode. In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). Always supported if lens is not fixed focus. Use android.lens.info.minimumFocusDistance to determine if lens is fixed-focus. Triggering AF_CANCEL resets the lens position to default, and sets the AF state to INACTIVE. MACRO Close-up focusing mode. In this mode, the lens does not move unless the autofocus trigger action is called. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). This mode is optimized for focusing on objects very close to the camera. When that trigger is activated, AF will transition to ACTIVE_SCAN, then to the outcome of the scan (FOCUSED or NOT_FOCUSED). Triggering cancel AF resets the lens position to default, and sets the AF state to INACTIVE. CONTINUOUS_VIDEO In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream. The focusing behavior should be suitable for good quality video recording; typically this means slower focus movement and no overshoots. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate. When the AF trigger is activated, the algorithm should immediately transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received. Once cancel is received, the algorithm should transition back to INACTIVE and resume passive scan. Note that this behavior is not identical to CONTINUOUS_PICTURE, since an ongoing PASSIVE_SCAN must immediately be canceled. CONTINUOUS_PICTURE In this mode, the AF algorithm modifies the lens position continually to attempt to provide a constantly-in-focus image stream. The focusing behavior should be suitable for still image capture; typically this means focusing as fast as possible. When the AF trigger is not involved, the AF algorithm should start in INACTIVE state, and then transition into PASSIVE_SCAN and PASSIVE_FOCUSED states as appropriate as it attempts to maintain focus. When the AF trigger is activated, the algorithm should finish its PASSIVE_SCAN if active, and then transition into AF_FOCUSED or AF_NOT_FOCUSED as appropriate, and lock the lens position until a cancel AF trigger is received. When the AF cancel trigger is activated, the algorithm should transition back to INACTIVE and then act as if it has just been started. EDOF Extended depth of field (digital focus) mode. The camera device will produce images with an extended depth of field automatically; no special focusing operations need to be done before taking a picture. AF triggers are ignored, and the AF state will always be INACTIVE. Whether auto-focus (AF) is currently enabled, and what mode it is set to. android.control.afAvailableModes

Only effective if android.control.mode = AUTO and the lens is not fixed focus (i.e. `android.lens.info.minimumFocusDistance > 0`). Also note that when android.control.aeMode is OFF, the behavior of AF is device dependent. It is recommended to lock AF by using android.control.afTrigger before setting android.control.aeMode to OFF, or set AF mode to OFF when AE is OFF. If the lens is controlled by the camera device auto-focus algorithm, the camera device will report the current AF status in android.control.afState in result metadata.

When afMode is AUTO or MACRO, the lens must not move until an AF trigger is sent in a request (android.control.afTrigger `==` START). After an AF trigger, the afState will end up with either FOCUSED_LOCKED or NOT_FOCUSED_LOCKED state (see android.control.afState for detailed state transitions), which indicates that the lens is locked and will not move. If camera movement (e.g. tilting camera) causes the lens to move after the lens is locked, the HAL must compensate this movement appropriately such that the same focal plane remains in focus. When afMode is one of the continuous auto focus modes, the HAL is free to start a AF scan whenever it's not locked. When the lens is locked after an AF trigger (see android.control.afState for detailed state transitions), the HAL should maintain the same lock behavior as above. When afMode is OFF, the application controls focus manually. The accuracy of the focus distance control depends on the android.lens.info.focusDistanceCalibration. However, the lens must not move regardless of the camera movement for any focus distance manual control. To put this in concrete terms, if the camera has lens elements which may move based on camera orientation or motion (e.g. due to gravity), then the HAL must drive the lens to remain in a fixed position invariant to the camera's orientation or motion, for example, by using accelerometer measurements in the lens control logic. This is a typical issue that will arise on camera modules with open-loop VCMs. 5 area_count List of metering areas to use for auto-focus. Pixel coordinates within android.sensor.info.activeArraySize Coordinates must be between `[(0,0), (width, height))` of android.sensor.info.activeArraySize

Not available if android.control.maxRegionsAf is 0. Otherwise will always be present. The maximum number of focus areas supported by the device is determined by the value of android.control.maxRegionsAf. The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array. The weight must be within `[0, 1000]`, and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region. The weights are relative to weights of other metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored. If all regions have 0 weight, then no specific metering area needs to be used by the camera device. If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax. IDLE The trigger is idle. START Autofocus will trigger now. CANCEL Autofocus will return to its initial state, and cancel any currently active trigger. Whether the camera device will trigger autofocus for this request.

This entry is normally set to IDLE, or is not included at all in the request settings. When included and set to START, the camera device will trigger the autofocus algorithm. If autofocus is disabled, this trigger has no effect. When set to CANCEL, the camera device will cancel any active trigger, and return to its initial AF state. Generally, applications should set this entry to START or CANCEL for only a single capture, and then return it to IDLE (or not set at all). Specifying START for multiple captures in a row means restarting the AF operation over and over again. See android.control.afState for what the trigger means for each AF mode.

OFF Auto-white balance lock is disabled; the AWB algorithm is free to update its parameters if in AUTO mode. ON Auto-white balance lock is enabled; the AWB algorithm will not update its parameters while the lock is active. Whether auto-white balance (AWB) is currently locked to its latest calculated values.

When set to `true` (ON), the AWB algorithm is locked to its latest parameters, and will not change color balance settings until the lock is set to `false` (OFF). Since the camera device has a pipeline of in-flight requests, the settings that get locked do not necessarily correspond to the settings that were present in the latest capture result received from the camera device, since additional captures and AWB updates may have occurred even before the result was sent out. If an application is switching between automatic and manual control and wishes to eliminate any flicker during the switch, the following procedure is recommended: 1. Starting in auto-AWB mode: 2. Lock AWB 3. Wait for the first result to be output that has the AWB locked 4. Copy AWB settings from that result into a request, set the request to manual AWB 5. Submit the capture request, proceed to run manual AWB as desired. Note that AWB lock is only meaningful when android.control.awbMode is in the AUTO mode; in other modes, AWB is already fixed to a specific setting. Some LEGACY devices may not support ON; the value is then overridden to OFF.

OFF The camera device's auto-white balance routine is disabled. The application-selected color transform matrix (android.colorCorrection.transform) and gains (android.colorCorrection.gains) are used by the camera device for manual white balance control. AUTO The camera device's auto-white balance routine is active. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. INCANDESCENT The camera device's auto-white balance routine is disabled; the camera device uses incandescent light as the assumed scene illumination for white balance. While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant A. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. FLUORESCENT The camera device's auto-white balance routine is disabled; the camera device uses fluorescent light as the assumed scene illumination for white balance. While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F2. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. WARM_FLUORESCENT The camera device's auto-white balance routine is disabled; the camera device uses warm fluorescent light as the assumed scene illumination for white balance. While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant F4. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. DAYLIGHT The camera device's auto-white balance routine is disabled; the camera device uses daylight light as the assumed scene illumination for white balance. While the exact white balance transforms are up to the camera device, they will approximately match the CIE standard illuminant D65. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. CLOUDY_DAYLIGHT The camera device's auto-white balance routine is disabled; the camera device uses cloudy daylight light as the assumed scene illumination for white balance. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. TWILIGHT The camera device's auto-white balance routine is disabled; the camera device uses twilight light as the assumed scene illumination for white balance. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. SHADE The camera device's auto-white balance routine is disabled; the camera device uses shade light as the assumed scene illumination for white balance. The application's values for android.colorCorrection.transform and android.colorCorrection.gains are ignored. For devices that support the MANUAL_POST_PROCESSING capability, the values used by the camera device for the transform and gains will be available in the capture result for this request. Whether auto-white balance (AWB) is currently setting the color transform fields, and what its illumination target is. android.control.awbAvailableModes

This control is only effective if android.control.mode is AUTO. When set to the ON mode, the camera device's auto-white balance routine is enabled, overriding the application's selected android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode. Note that when android.control.aeMode is OFF, the behavior of AWB is device dependent. It is recommened to also set AWB mode to OFF or lock AWB by using android.control.awbLock before setting AE mode to OFF. When set to the OFF mode, the camera device's auto-white balance routine is disabled. The application manually controls the white balance by android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode. When set to any other modes, the camera device's auto-white balance routine is disabled. The camera device uses each particular illumination target for white balance adjustment. The application's values for android.colorCorrection.transform, android.colorCorrection.gains and android.colorCorrection.mode are ignored.

5 area_count List of metering areas to use for auto-white-balance illuminant estimation. Pixel coordinates within android.sensor.info.activeArraySize Coordinates must be between `[(0,0), (width, height))` of android.sensor.info.activeArraySize

Not available if android.control.maxRegionsAwb is 0. Otherwise will always be present. The maximum number of regions supported by the device is determined by the value of android.control.maxRegionsAwb. The coordinate system is based on the active pixel array, with (0,0) being the top-left pixel in the active pixel array, and (android.sensor.info.activeArraySize.width - 1, android.sensor.info.activeArraySize.height - 1) being the bottom-right pixel in the active pixel array. The weight must range from 0 to 1000, and represents a weight for every pixel in the area. This means that a large metering area with the same weight as a smaller area will have more effect in the metering result. Metering areas can partially overlap and the camera device will add the weights in the overlap region. The weights are relative to weights of other white balance metering regions, so if only one region is used, all non-zero weights will have the same effect. A region with 0 weight is ignored. If all regions have 0 weight, then no specific metering area needs to be used by the camera device. If the metering region is outside the used android.scaler.cropRegion returned in capture result metadata, the camera device will ignore the sections outside the crop region and output only the intersection rectangle as the metering region in the result metadata. If the region is entirely outside the crop region, it will be ignored and not reported in the result metadata.

The HAL level representation of MeteringRectangle[] is a int[5 * area_count]. Every five elements represent a metering region of (xmin, ymin, xmax, ymax, weight). The rectangle is defined to be inclusive on xmin and ymin, but exclusive on xmax and ymax. CUSTOM The goal of this request doesn't fall into the other categories. The camera device will default to preview-like behavior. PREVIEW This request is for a preview-like use case. The precapture trigger may be used to start off a metering w/flash sequence. STILL_CAPTURE This request is for a still capture-type use case. If the flash unit is under automatic control, it may fire as needed. VIDEO_RECORD This request is for a video recording use case. VIDEO_SNAPSHOT This request is for a video snapshot (still image while recording video) use case. The camera device should take the highest-quality image possible (given the other settings) without disrupting the frame rate of video recording. ZERO_SHUTTER_LAG This request is for a ZSL usecase; the application will stream full-resolution images and reprocess one or several later for a final capture. MANUAL This request is for manual capture use case where the applications want to directly control the capture parameters. For example, the application may wish to manually control android.sensor.exposureTime, android.sensor.sensitivity, etc. Information to the camera device 3A (auto-exposure, auto-focus, auto-white balance) routines about the purpose of this capture, to help the camera device to decide optimal 3A strategy.

This control (except for MANUAL) is only effective if `android.control.mode != OFF` and any 3A routine is active. ZERO_SHUTTER_LAG will be supported if android.request.availableCapabilities contains PRIVATE_REPROCESSING or YUV_REPROCESSING. MANUAL will be supported if android.request.availableCapabilities contains MANUAL_SENSOR. Other intent values are always supported.

OFF No color effect will be applied. MONO A "monocolor" effect where the image is mapped into a single color. This will typically be grayscale. NEGATIVE A "photo-negative" effect where the image's colors are inverted. SOLARIZE A "solarisation" effect (Sabattier effect) where the image is wholly or partially reversed in tone. SEPIA A "sepia" effect where the image is mapped into warm gray, red, and brown tones. POSTERIZE A "posterization" effect where the image uses discrete regions of tone rather than a continuous gradient of tones. WHITEBOARD A "whiteboard" effect where the image is typically displayed as regions of white, with black or grey details. BLACKBOARD A "blackboard" effect where the image is typically displayed as regions of black, with white or grey details. AQUA An "aqua" effect where a blue hue is added to the image. A special color effect to apply. android.control.availableEffects

When this mode is set, a color effect will be applied to images produced by the camera device. The interpretation and implementation of these color effects is left to the implementor of the camera device, and should not be depended on to be consistent (or present) across all devices.

OFF Full application control of pipeline. All control by the device's metering and focusing (3A) routines is disabled, and no other settings in android.control.* have any effect, except that android.control.captureIntent may be used by the camera device to select post-processing values for processing blocks that do not allow for manual control, or are not exposed by the camera API. However, the camera device's 3A routines may continue to collect statistics and update their internal state so that when control is switched to AUTO mode, good control values can be immediately applied. AUTO Use settings for each individual 3A routine. Manual control of capture parameters is disabled. All controls in android.control.* besides sceneMode take effect. USE_SCENE_MODE Use a specific scene mode. Enabling this disables control.aeMode, control.awbMode and control.afMode controls; the camera device will ignore those settings while USE_SCENE_MODE is active (except for FACE_PRIORITY scene mode). Other control entries are still active. This setting can only be used if scene mode is supported (i.e. android.control.availableSceneModes contain some modes other than DISABLED). OFF_KEEP_STATE Same as OFF mode, except that this capture will not be used by camera device background auto-exposure, auto-white balance and auto-focus algorithms (3A) to update their statistics. Specifically, the 3A routines are locked to the last values set from a request with AUTO, OFF, or USE_SCENE_MODE, and any statistics or state updates collected from manual captures with OFF_KEEP_STATE will be discarded by the camera device. Overall mode of 3A (auto-exposure, auto-white-balance, auto-focus) control routines. android.control.availableModes

This is a top-level 3A control switch. When set to OFF, all 3A control by the camera device is disabled. The application must set the fields for capture parameters itself. When set to AUTO, the individual algorithm controls in android.control.* are in effect, such as android.control.afMode. When set to USE_SCENE_MODE, the individual controls in android.control.* are mostly disabled, and the camera device implements one of the scene mode settings (such as ACTION, SUNSET, or PARTY) as it wishes. The camera device scene mode 3A settings are provided by {@link android.hardware.camera2.CaptureResult capture results}. When set to OFF_KEEP_STATE, it is similar to OFF mode, the only difference is that this frame will not be used by camera device background 3A statistics update, as if this frame is never captured. This mode can be used in the scenario where the application doesn't want a 3A manual control capture to affect the subsequent auto 3A capture results.

DISABLED Indicates that no scene modes are set for a given capture request. FACE_PRIORITY If face detection support exists, use face detection data for auto-focus, auto-white balance, and auto-exposure routines. If face detection statistics are disabled (i.e. android.statistics.faceDetectMode is set to OFF), this should still operate correctly (but will not return face detection statistics to the framework). Unlike the other scene modes, android.control.aeMode, android.control.awbMode, and android.control.afMode remain active when FACE_PRIORITY is set. ACTION Optimized for photos of quickly moving objects. Similar to SPORTS. PORTRAIT Optimized for still photos of people. LANDSCAPE Optimized for photos of distant macroscopic objects. NIGHT Optimized for low-light settings. NIGHT_PORTRAIT Optimized for still photos of people in low-light settings. THEATRE Optimized for dim, indoor settings where flash must remain off. BEACH Optimized for bright, outdoor beach settings. SNOW Optimized for bright, outdoor settings containing snow. SUNSET Optimized for scenes of the setting sun. STEADYPHOTO Optimized to avoid blurry photos due to small amounts of device motion (for example: due to hand shake). FIREWORKS Optimized for nighttime photos of fireworks. SPORTS Optimized for photos of quickly moving people. Similar to ACTION. PARTY Optimized for dim, indoor settings with multiple moving people. CANDLELIGHT Optimized for dim settings where the main light source is a flame. BARCODE Optimized for accurately capturing a photo of barcode for use by camera applications that wish to read the barcode value. HIGH_SPEED_VIDEO This is deprecated, please use {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession} and {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedRequestList} for high speed video recording. Optimized for high speed video recording (frame rate >=60fps) use case. The supported high speed video sizes and fps ranges are specified in android.control.availableHighSpeedVideoConfigurations. To get desired output frame rates, the application is only allowed to select video size and fps range combinations listed in this static metadata. The fps range can be control via android.control.aeTargetFpsRange. In this mode, the camera device will override aeMode, awbMode, and afMode to ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode controls will be overridden to be FAST. Therefore, no manual control of capture and post-processing parameters is possible. All other controls operate the same as when android.control.mode == AUTO. This means that all other android.control.* fields continue to work, such as * android.control.aeTargetFpsRange * android.control.aeExposureCompensation * android.control.aeLock * android.control.awbLock * android.control.effectMode * android.control.aeRegions * android.control.afRegions * android.control.awbRegions * android.control.afTrigger * android.control.aePrecaptureTrigger Outside of android.control.*, the following controls will work: * android.flash.mode (automatic flash for still capture will not work since aeMode is ON) * android.lens.opticalStabilizationMode (if it is supported) * android.scaler.cropRegion * android.statistics.faceDetectMode For high speed recording use case, the actual maximum supported frame rate may be lower than what camera can output, depending on the destination Surfaces for the image data. For example, if the destination surface is from video encoder, the application need check if the video encoder is capable of supporting the high frame rate for a given video size, or it will end up with lower recording frame rate. If the destination surface is from preview window, the preview frame rate will be bounded by the screen refresh rate. The camera device will only support up to 2 output high speed streams (processed non-stalling format defined in android.request.maxNumOutputStreams) in this mode. This control will be effective only if all of below conditions are true: * The application created no more than maxNumHighSpeedStreams processed non-stalling format output streams, where maxNumHighSpeedStreams is calculated as min(2, android.request.maxNumOutputStreams[Processed (but not-stalling)]). * The stream sizes are selected from the sizes reported by android.control.availableHighSpeedVideoConfigurations. * No processed non-stalling or raw streams are configured. When above conditions are NOT satistied, the controls of this mode and android.control.aeTargetFpsRange will be ignored by the camera device, the camera device will fall back to android.control.mode `==` AUTO, and the returned capture result metadata will give the fps range choosen by the camera device. Switching into or out of this mode may trigger some camera ISP/sensor reconfigurations, which may introduce extra latency. It is recommended that the application avoids unnecessary scene mode switch as much as possible. HDR Turn on a device-specific high dynamic range (HDR) mode. In this scene mode, the camera device captures images that keep a larger range of scene illumination levels visible in the final image. For example, when taking a picture of a object in front of a bright window, both the object and the scene through the window may be visible when using HDR mode, while in normal AUTO mode, one or the other may be poorly exposed. As a tradeoff, HDR mode generally takes much longer to capture a single image, has no user control, and may have other artifacts depending on the HDR method used. Therefore, HDR captures operate at a much slower rate than regular captures. In this mode, on LIMITED or FULL devices, when a request is made with a android.control.captureIntent of STILL_CAPTURE, the camera device will capture an image using a high dynamic range capture technique. On LEGACY devices, captures that target a JPEG-format output will be captured with HDR, and the capture intent is not relevant. The HDR capture may involve the device capturing a burst of images internally and combining them into one, or it may involve the device using specialized high dynamic range capture hardware. In all cases, a single image is produced in response to a capture request submitted while in HDR mode. Since substantial post-processing is generally needed to produce an HDR image, only YUV and JPEG outputs are supported for LIMITED/FULL device HDR captures, and only JPEG outputs are supported for LEGACY HDR captures. Using a RAW output for HDR capture is not supported. Control for which scene mode is currently active. android.control.availableSceneModes

Scene modes are custom camera modes optimized for a certain set of conditions and capture settings. This is the mode that that is active when `android.control.mode == USE_SCENE_MODE`. Aside from FACE_PRIORITY, these modes will disable android.control.aeMode, android.control.awbMode, and android.control.afMode while in use. The interpretation and implementation of these scene modes is left to the implementor of the camera device. Their behavior will not be consistent across all devices, and any given device may only implement a subset of these modes.

HAL implementations that include scene modes are expected to provide the per-scene settings to use for android.control.aeMode, android.control.awbMode, and android.control.afMode in android.control.sceneModeOverrides. For HIGH_SPEED_VIDEO mode, if it is included in android.control.availableSceneModes, the HAL must list supported video size and fps range in android.control.availableHighSpeedVideoConfigurations. For a given size, e.g. 1280x720, if the HAL has two different sensor configurations for normal streaming mode and high speed streaming, when this scene mode is set/reset in a sequence of capture requests, the HAL may have to switch between different sensor modes. This mode is deprecated in HAL3.3, to support high speed video recording, please implement android.control.availableHighSpeedVideoConfigurations and CONSTRAINED_HIGH_SPEED_VIDEO capbility defined in android.request.availableCapabilities. OFF Video stabilization is disabled. ON Video stabilization is enabled. Whether video stabilization is active.

Video stabilization automatically translates and scales images from the camera in order to stabilize motion between consecutive frames. If enabled, video stabilization can modify the android.scaler.cropRegion to keep the video stream stabilized. Switching between different video stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the video stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done. If a camera device supports both this mode and OIS (android.lens.opticalStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time.

n List of auto-exposure antibanding modes for android.control.aeAntibandingMode that are supported by this camera device. Any value listed in android.control.aeAntibandingMode

Not all of the auto-exposure anti-banding modes may be supported by a given camera device. This field lists the valid anti-banding modes that the application may request for this camera device with the android.control.aeAntibandingMode control.

n List of auto-exposure modes for android.control.aeMode that are supported by this camera device. Any value listed in android.control.aeMode

Not all the auto-exposure modes may be supported by a given camera device, especially if no flash unit is available. This entry lists the valid modes for android.control.aeMode for this camera device. All camera devices support ON, and all camera devices with flash units support ON_AUTO_FLASH and ON_ALWAYS_FLASH. FULL mode camera devices always support OFF mode, which enables application control of camera exposure time, sensitivity, and frame duration. LEGACY mode camera devices never support OFF mode. LIMITED mode devices support OFF if they support the MANUAL_SENSOR capability.

2 n List of frame rate ranges for android.control.aeTargetFpsRange supported by this camera device. Frames per second (FPS)

For devices at the LEGACY level or above: * This list will always include (30, 30). * Also, for constant-framerate recording, for each normal {@link android.media.CamcorderProfile CamcorderProfile} that has {@link android.media.CamcorderProfile#quality quality} in the range [{@link android.media.CamcorderProfile#QUALITY_LOW QUALITY_LOW}, {@link android.media.CamcorderProfile#QUALITY_2160P QUALITY_2160P}], if the profile is supported by the device and has {@link android.media.CamcorderProfile#videoFrameRate videoFrameRate} `x`, this list will always include (`x`,`x`). * For preview streaming use case, this list will always include (`min`, `max`) where `min` <= 15 and `max` >= 30. For devices at the LIMITED level or above: * For YUV_420_888 burst capture use case, this list will always include (`min`, `max`) and (`max`, `max`) where `min` <= 15 and `max` = the maximum output frame rate of the maximum YUV_420_888 output size.

2 Maximum and minimum exposure compensation values for android.control.aeExposureCompensation, in counts of android.control.aeCompensationStep, that are supported by this camera device. Range [0,0] indicates that exposure compensation is not supported. For LIMITED and FULL devices, range must follow below requirements if exposure compensation is supported (`range != [0, 0]`): `Min.exposure compensation * android.control.aeCompensationStep <= -2 EV` `Max.exposure compensation * android.control.aeCompensationStep >= 2 EV` LEGACY devices may support a smaller range than this. Smallest step by which the exposure compensation can be changed. Exposure Value (EV)

This is the unit for android.control.aeExposureCompensation. For example, if this key has a value of `1/2`, then a setting of `-2` for android.control.aeExposureCompensation means that the target EV offset for the auto-exposure routine is -1 EV. One unit of EV compensation changes the brightness of the captured image by a factor of two. +1 EV doubles the image brightness, while -1 EV halves the image brightness.

This must be less than or equal to 1/2. n List of auto-focus (AF) modes for android.control.afMode that are supported by this camera device. Any value listed in android.control.afMode

Not all the auto-focus modes may be supported by a given camera device. This entry lists the valid modes for android.control.afMode for this camera device. All LIMITED and FULL mode camera devices will support OFF mode, and all camera devices with adjustable focuser units (`android.lens.info.minimumFocusDistance > 0`) will support AUTO mode. LEGACY devices will support OFF mode only if they support focusing to infinity (by also setting android.lens.focusDistance to `0.0f`).

n List of color effects for android.control.effectMode that are supported by this camera device. Any value listed in android.control.effectMode

This list contains the color effect modes that can be applied to images produced by the camera device. Implementations are not expected to be consistent across all devices. If no color effect modes are available for a device, this will only list OFF. A color effect will only be applied if android.control.mode != OFF. OFF is always included in this list. This control has no effect on the operation of other control routines such as auto-exposure, white balance, or focus.

n List of scene modes for android.control.sceneMode that are supported by this camera device. Any value listed in android.control.sceneMode

This list contains scene modes that can be set for the camera device. Only scene modes that have been fully implemented for the camera device may be included here. Implementations are not expected to be consistent across all devices. If no scene modes are supported by the camera device, this will be set to DISABLED. Otherwise DISABLED will not be listed. FACE_PRIORITY is always listed if face detection is supported (i.e.`android.statistics.info.maxFaceCount > 0`).

n List of video stabilization modes for android.control.videoStabilizationMode that are supported by this camera device. Any value listed in android.control.videoStabilizationMode

OFF will always be listed.

n List of auto-white-balance modes for android.control.awbMode that are supported by this camera device. Any value listed in android.control.awbMode

Not all the auto-white-balance modes may be supported by a given camera device. This entry lists the valid modes for android.control.awbMode for this camera device. All camera devices will support ON mode. Camera devices that support the MANUAL_POST_PROCESSING capability will always support OFF mode, which enables application control of white balance, by using android.colorCorrection.transform and android.colorCorrection.gains (android.colorCorrection.mode must be set to TRANSFORM_MATRIX). This includes all FULL mode camera devices.

3 List of the maximum number of regions that can be used for metering in auto-exposure (AE), auto-white balance (AWB), and auto-focus (AF); this corresponds to the the maximum number of elements in android.control.aeRegions, android.control.awbRegions, and android.control.afRegions. Value must be >= 0 for each element. For full-capability devices this value must be >= 1 for AE and AF. The order of the elements is: `(AE, AWB, AF)`. The maximum number of metering regions that can be used by the auto-exposure (AE) routine. Value will be >= 0. For FULL-capability devices, this value will be >= 1.

This corresponds to the the maximum allowed number of elements in android.control.aeRegions.

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated. The maximum number of metering regions that can be used by the auto-white balance (AWB) routine. Value will be >= 0.

This corresponds to the the maximum allowed number of elements in android.control.awbRegions.

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated. The maximum number of metering regions that can be used by the auto-focus (AF) routine. Value will be >= 0. For FULL-capability devices, this value will be >= 1.

This corresponds to the the maximum allowed number of elements in android.control.afRegions.

This entry is private to the framework. Fill in maxRegions to have this entry be automatically populated. 3 length(availableSceneModes) Ordered list of auto-exposure, auto-white balance, and auto-focus settings to use with each available scene mode. For each available scene mode, the list must contain three entries containing the android.control.aeMode, android.control.awbMode, and android.control.afMode values used by the camera device. The entry order is `(aeMode, awbMode, afMode)` where aeMode has the lowest index position.

When a scene mode is enabled, the camera device is expected to override android.control.aeMode, android.control.awbMode, and android.control.afMode with its preferred settings for that scene mode. The order of this list matches that of availableSceneModes, with 3 entries for each mode. The overrides listed for FACE_PRIORITY are ignored, since for that mode the application-set android.control.aeMode, android.control.awbMode, and android.control.afMode values are used instead, matching the behavior when android.control.mode is set to AUTO. It is recommended that the FACE_PRIORITY overrides should be set to 0. For example, if availableSceneModes contains `(FACE_PRIORITY, ACTION, NIGHT)`, then the camera framework expects sceneModeOverrides to have 9 entries formatted like: `(0, 0, 0, ON_AUTO_FLASH, AUTO, CONTINUOUS_PICTURE, ON_AUTO_FLASH, INCANDESCENT, AUTO)`.

To maintain backward compatibility, this list will be made available in the static metadata of the camera service. The camera service will use these values to set android.control.aeMode, android.control.awbMode, and android.control.afMode when using a scene mode other than FACE_PRIORITY. The ID sent with the latest CAMERA2_TRIGGER_PRECAPTURE_METERING call

Must be 0 if no CAMERA2_TRIGGER_PRECAPTURE_METERING trigger received yet by HAL. Always updated even if AE algorithm ignores the trigger

INACTIVE AE is off or recently reset. When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result. SEARCHING AE doesn't yet have a good set of control values for the current scene. This is a transient state, the camera device may skip reporting this state in capture result. CONVERGED AE has a good set of control values for the current scene. LOCKED AE has been locked. FLASH_REQUIRED AE has a good set of control values, but flash needs to be fired for good quality still capture. PRECAPTURE AE has been asked to do a precapture sequence and is currently executing it. Precapture can be triggered through setting android.control.aePrecaptureTrigger to START. Currently active and completed (if it causes camera device internal AE lock) precapture metering sequence can be canceled through setting android.control.aePrecaptureTrigger to CANCEL. Once PRECAPTURE completes, AE will transition to CONVERGED or FLASH_REQUIRED as appropriate. This is a transient state, the camera device may skip reporting this state in capture result. Current state of the auto-exposure (AE) algorithm.

Switching between or enabling AE modes (android.control.aeMode) always resets the AE state to INACTIVE. Similarly, switching between android.control.mode, or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all the algorithm states to INACTIVE. The camera device can do several state transitions between two results, if it is allowed by the state transition table. For example: INACTIVE may never actually be seen in a result. The state in the result is the state for this image (in sync with this image): if AE state becomes CONVERGED, then the image data associated with this result should be good to use. Below are state transition tables for different AE modes. State | Transition Cause | New State | Notes :------------:|:----------------:|:---------:|:-----------------------: INACTIVE | | INACTIVE | Camera device auto exposure algorithm is disabled When android.control.aeMode is AE_MODE_ON_*: State | Transition Cause | New State | Notes :-------------:|:--------------------------------------------:|:--------------:|:-----------------: INACTIVE | Camera device initiates AE scan | SEARCHING | Values changing INACTIVE | android.control.aeLock is ON | LOCKED | Values locked SEARCHING | Camera device finishes AE scan | CONVERGED | Good values, not changing SEARCHING | Camera device finishes AE scan | FLASH_REQUIRED | Converged but too dark w/o flash SEARCHING | android.control.aeLock is ON | LOCKED | Values locked CONVERGED | Camera device initiates AE scan | SEARCHING | Values changing CONVERGED | android.control.aeLock is ON | LOCKED | Values locked FLASH_REQUIRED | Camera device initiates AE scan | SEARCHING | Values changing FLASH_REQUIRED | android.control.aeLock is ON | LOCKED | Values locked LOCKED | android.control.aeLock is OFF | SEARCHING | Values not good after unlock LOCKED | android.control.aeLock is OFF | CONVERGED | Values good after unlock LOCKED | android.control.aeLock is OFF | FLASH_REQUIRED | Exposure good, but too dark PRECAPTURE | Sequence done. android.control.aeLock is OFF | CONVERGED | Ready for high-quality capture PRECAPTURE | Sequence done. android.control.aeLock is ON | LOCKED | Ready for high-quality capture LOCKED | aeLock is ON and aePrecaptureTrigger is START | LOCKED | Precapture trigger is ignored when AE is already locked LOCKED | aeLock is ON and aePrecaptureTrigger is CANCEL| LOCKED | Precapture trigger is ignored when AE is already locked Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START | PRECAPTURE | Start AE precapture metering sequence Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL| INACTIVE | Currently active precapture metering sequence is canceled For the above table, the camera device may skip reporting any state changes that happen without application intervention (i.e. mode switch, trigger, locking). Any state that can be skipped in that manner is called a transient state. For example, for above AE modes (AE_MODE_ON_*), in addition to the state transitions listed in above table, it is also legal for the camera device to skip one or more transient states between two results. See below table for examples: State | Transition Cause | New State | Notes :-------------:|:-----------------------------------------------------------:|:--------------:|:-----------------: INACTIVE | Camera device finished AE scan | CONVERGED | Values are already good, transient states are skipped by camera device. Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START, sequence done | FLASH_REQUIRED | Converged but too dark w/o flash after a precapture sequence, transient states are skipped by camera device. Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is START, sequence done | CONVERGED | Converged after a precapture sequence, transient states are skipped by camera device. Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL, converged | FLASH_REQUIRED | Converged but too dark w/o flash after a precapture sequence is canceled, transient states are skipped by camera device. Any state (excluding LOCKED) | android.control.aePrecaptureTrigger is CANCEL, converged | CONVERGED | Converged after a precapture sequenceis canceled, transient states are skipped by camera device. CONVERGED | Camera device finished AE scan | FLASH_REQUIRED | Converged but too dark w/o flash after a new scan, transient states are skipped by camera device. FLASH_REQUIRED | Camera device finished AE scan | CONVERGED | Converged after a new scan, transient states are skipped by camera device.

INACTIVE AF is off or has not yet tried to scan/been asked to scan. When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result. PASSIVE_SCAN AF is currently performing an AF scan initiated the camera device in a continuous autofocus mode. Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result. PASSIVE_FOCUSED AF currently believes it is in focus, but may restart scanning at any time. Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result. ACTIVE_SCAN AF is performing an AF scan because it was triggered by AF trigger. Only used by AUTO or MACRO AF modes. This is a transient state, the camera device may skip reporting this state in capture result. FOCUSED_LOCKED AF believes it is focused correctly and has locked focus. This state is reached only after an explicit START AF trigger has been sent (android.control.afTrigger), when good focus has been obtained. The lens will remain stationary until the AF mode (android.control.afMode) is changed or a new AF trigger is sent to the camera device (android.control.afTrigger). NOT_FOCUSED_LOCKED AF has failed to focus successfully and has locked focus. This state is reached only after an explicit START AF trigger has been sent (android.control.afTrigger), when good focus cannot be obtained. The lens will remain stationary until the AF mode (android.control.afMode) is changed or a new AF trigger is sent to the camera device (android.control.afTrigger). PASSIVE_UNFOCUSED AF finished a passive scan without finding focus, and may restart scanning at any time. Only used by CONTINUOUS_* AF modes. This is a transient state, the camera device may skip reporting this state in capture result. LEGACY camera devices do not support this state. When a passive scan has finished, it will always go to PASSIVE_FOCUSED. Current state of auto-focus (AF) algorithm.

The ID sent with the latest CAMERA2_TRIGGER_AUTOFOCUS call

Must be 0 if no CAMERA2_TRIGGER_AUTOFOCUS trigger received yet by HAL. Always updated even if AF algorithm ignores the trigger

INACTIVE AWB is not in auto mode, or has not yet started metering. When a camera device is opened, it starts in this state. This is a transient state, the camera device may skip reporting this state in capture result. SEARCHING AWB doesn't yet have a good set of control values for the current scene. This is a transient state, the camera device may skip reporting this state in capture result. CONVERGED AWB has a good set of control values for the current scene. LOCKED AWB has been locked. Current state of auto-white balance (AWB) algorithm.

Switching between or enabling AWB modes (android.control.awbMode) always resets the AWB state to INACTIVE. Similarly, switching between android.control.mode, or android.control.sceneMode if `android.control.mode == USE_SCENE_MODE` resets all the algorithm states to INACTIVE. The camera device can do several state transitions between two results, if it is allowed by the state transition table. So INACTIVE may never actually be seen in a result. The state in the result is the state for this image (in sync with this image): if AWB state becomes CONVERGED, then the image data associated with this result should be good to use. Below are state transition tables for different AWB modes. When `android.control.awbMode != AWB_MODE_AUTO`: State | Transition Cause | New State | Notes :------------:|:----------------:|:---------:|:-----------------------: INACTIVE | |INACTIVE |Camera device auto white balance algorithm is disabled When android.control.awbMode is AWB_MODE_AUTO: State | Transition Cause | New State | Notes :-------------:|:--------------------------------:|:-------------:|:-----------------: INACTIVE | Camera device initiates AWB scan | SEARCHING | Values changing INACTIVE | android.control.awbLock is ON | LOCKED | Values locked SEARCHING | Camera device finishes AWB scan | CONVERGED | Good values, not changing SEARCHING | android.control.awbLock is ON | LOCKED | Values locked CONVERGED | Camera device initiates AWB scan | SEARCHING | Values changing CONVERGED | android.control.awbLock is ON | LOCKED | Values locked LOCKED | android.control.awbLock is OFF | SEARCHING | Values not good after unlock For the above table, the camera device may skip reporting any state changes that happen without application intervention (i.e. mode switch, trigger, locking). Any state that can be skipped in that manner is called a transient state. For example, for this AWB mode (AWB_MODE_AUTO), in addition to the state transitions listed in above table, it is also legal for the camera device to skip one or more transient states between two results. See below table for examples: State | Transition Cause | New State | Notes :-------------:|:--------------------------------:|:-------------:|:-----------------: INACTIVE | Camera device finished AWB scan | CONVERGED | Values are already good, transient states are skipped by camera device. LOCKED | android.control.awbLock is OFF | CONVERGED | Values good after unlock, transient states are skipped by camera device.

5 n List of available high speed video size, fps range and max batch size configurations supported by the camera device, in the format of (width, height, fps_min, fps_max, batch_size_max). For each configuration, the fps_max >= 120fps.

When CONSTRAINED_HIGH_SPEED_VIDEO is supported in android.control.availableCapabilities, this metadata will list the supported high speed video size, fps range and max batch size configurations. All the sizes listed in this configuration will be a subset of the sizes reported by {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes} for processed non-stalling formats. For the high speed video use case, the application must select the video size and fps range from this metadata to configure the recording and preview streams and setup the recording requests. For example, if the application intends to do high speed recording, it can select the maximum size reported by this metadata to configure output streams. Once the size is selected, application can filter this metadata by selected size and get the supported fps ranges, and use these fps ranges to setup the recording requests. Note that for the use case of multiple output streams, application must select one unique size from this metadata to use (e.g., preview and recording streams must have the same size). Otherwise, the high speed capture session creation will fail. The min and max fps will be multiple times of 30fps. High speed video streaming extends significant performance pressue to camera hardware, to achieve efficient high speed streaming, the camera device may have to aggregate multiple frames together and send to camera device for processing where the request controls are same for all the frames in this batch. Max batch size indicates the max possible number of frames the camera device will group together for this high speed stream configuration. This max batch size will be used to generate a high speed recording request list by {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedRequestList}. The max batch size for each configuration will satisfy below conditions: * Each max batch size will be a divisor of its corresponding fps_max / 30. For example, if max_fps is 300, max batch size will only be 1, 2, 5, or 10. * The camera device may choose smaller internal batch size for each configuration, but the actual batch size will be a divisor of max batch size. For example, if the max batch size is 8, the actual batch size used by camera device will only be 1, 2, 4, or 8. * The max batch size in each configuration entry must be no larger than 32. The camera device doesn't have to support batch mode to achieve high speed video recording, in such case, batch_size_max will be reported as 1 in each configuration entry. This fps ranges in this configuration list can only be used to create requests that are submitted to a high speed camera capture session created by {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}. The fps ranges reported in this metadata must not be used to setup capture requests for normal capture session, or it will cause request error.

All the sizes listed in this configuration will be a subset of the sizes reported by android.scaler.availableStreamConfigurations for processed non-stalling output formats. Note that for all high speed video configurations, HAL must be able to support a minimum of two streams, though the application might choose to configure just one stream. The HAL may support multiple sensor modes for high speed outputs, for example, 120fps sensor mode and 120fps recording, 240fps sensor mode for 240fps recording. The application usually starts preview first, then starts recording. To avoid sensor mode switch caused stutter when starting recording as much as possible, the application may want to ensure the same sensor mode is used for preview and recording. Therefore, The HAL must advertise the variable fps range [30, fps_max] for each fixed fps range in this configuration list. For example, if the HAL advertises [120, 120] and [240, 240], the HAL must also advertise [30, 120] and [30, 240] for each configuration. In doing so, if the application intends to do 120fps recording, it can select [30, 120] to start preview, and [120, 120] to start recording. For these variable fps ranges, it's up to the HAL to decide the actual fps values that are suitable for smooth preview streaming. If the HAL sees different max_fps values that fall into different sensor modes in a sequence of requests, the HAL must switch the sensor mode as quick as possible to minimize the mode switch caused stutter. FALSE TRUE Whether the camera device supports android.control.aeLock

Devices with MANUAL_SENSOR capability or BURST_CAPTURE capability will always list `true`. This includes FULL devices.

FALSE TRUE Whether the camera device supports android.control.awbLock

Devices with MANUAL_POST_PROCESSING capability or BURST_CAPTURE capability will always list `true`. This includes FULL devices.

n List of control modes for android.control.mode that are supported by this camera device. Any value listed in android.control.mode

This list contains control modes that can be set for the camera device. LEGACY mode devices will always support AUTO mode. LIMITED and FULL devices will always support OFF, AUTO modes.

FAST Minimal or no slowdown of frame rate compared to Bayer RAW output. HIGH_QUALITY Improved processing quality but the frame rate might be slowed down relative to raw output. Controls the quality of the demosaicing processing.

OFF No edge enhancement is applied. FAST Apply edge enhancement at a quality level that does not slow down frame rate relative to sensor output HIGH_QUALITY Apply high-quality edge enhancement, at a cost of possibly reduced output frame rate. Operation mode for edge enhancement. android.edge.availableEdgeModes

Edge enhancement improves sharpness and details in the captured image. OFF means no enhancement will be applied by the camera device. FAST/HIGH_QUALITY both mean camera device determined enhancement will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality enhancement algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying edge enhancement. For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV-domain edge enhancement, respectively. The camera device may adjust its internal noise reduction parameters for best image quality based on the android.reprocess.effectiveExposureFactor, if it is set.

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal edge enhancement reduction parameters appropriately to get the best quality images. Control the amount of edge enhancement applied to the images 1-10; 10 is maximum sharpening n List of edge enhancement modes for android.edge.mode that are supported by this camera device. Any value listed in android.edge.mode

Full-capability camera devices must always support OFF; all devices will list FAST.

HAL must support both FAST and HIGH_QUALITY if edge enhancement control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

Power for flash firing/torch 10 is max power; 0 is no flash. Linear 0 - 10

Power for snapshot may use a different scale than for torch mode. Only one entry for torch mode will be used

Firing time of flash relative to start of exposure nanoseconds 0-(exposure time-flash duration)

Clamped to (0, exposure time - flash duration).

OFF Do not fire the flash for this capture. SINGLE If the flash is available and charged, fire flash for this capture. TORCH Transition flash to continuously on. The desired mode for for the camera device's flash control.

This control is only effective when flash unit is available (`android.flash.info.available == true`). When this control is used, the android.control.aeMode must be set to ON or OFF. Otherwise, the camera device auto-exposure related flash control (ON_AUTO_FLASH, ON_ALWAYS_FLASH, or ON_AUTO_FLASH_REDEYE) will override this control. When set to OFF, the camera device will not fire flash for this capture. When set to SINGLE, the camera device will fire flash regardless of the camera device's auto-exposure routine's result. When used in still capture case, this control should be used along with auto-exposure (AE) precapture metering sequence (android.control.aePrecaptureTrigger), otherwise, the image may be incorrectly exposed. When set to TORCH, the flash will be on continuously. This mode can be used for use cases such as preview, auto-focus assist, still capture, or video recording. The flash status will be reported by android.flash.state in the capture result metadata.

FALSE TRUE Whether this camera device has a flash unit.

Will be `false` if no flash is available. If there is no flash unit, none of the flash controls do anything.

Time taken before flash can fire again nanoseconds 0-1e9

1 second too long/too short for recharge? Should this be power-dependent?

The x,y whitepoint of the flash pair of floats 0-1 for both Max energy output of the flash for a full power single flash lumen-seconds >= 0 UNAVAILABLE No flash on camera. CHARGING Flash is charging and cannot be fired. READY Flash is ready to fire. FIRED Flash fired for this capture. PARTIAL Flash partially illuminated this frame. This is usually due to the next or previous frame having the flash fire, and the flash spilling into this capture due to hardware limitations. Current state of the flash unit.

When the camera device doesn't have flash unit (i.e. `android.flash.info.available == false`), this state will always be UNAVAILABLE. Other states indicate the current flash status. In certain conditions, this will be available on LEGACY devices: * Flash-less cameras always return UNAVAILABLE. * Using android.control.aeMode `==` ON_ALWAYS_FLASH will always return FIRED. * Using android.flash.mode `==` TORCH will always return FIRED. In all other conditions the state will not be available on LEGACY devices (i.e. it will be `null`).

OFF No hot pixel correction is applied. The frame rate must not be reduced relative to sensor raw output for this option. The hotpixel map may be returned in android.statistics.hotPixelMap. FAST Hot pixel correction is applied, without reducing frame rate relative to sensor raw output. The hotpixel map may be returned in android.statistics.hotPixelMap. HIGH_QUALITY High-quality hot pixel correction is applied, at a cost of possibly reduced frame rate relative to sensor raw output. The hotpixel map may be returned in android.statistics.hotPixelMap. Operational mode for hot pixel correction. android.hotPixel.availableHotPixelModes

Hotpixel correction interpolates out, or otherwise removes, pixels that do not accurately measure the incoming light (i.e. pixels that are stuck at an arbitrary value or are oversensitive).

n List of hot pixel correction modes for android.hotPixel.mode that are supported by this camera device. Any value listed in android.hotPixel.mode

FULL mode camera devices will always support FAST.

To avoid performance issues, there will be significantly fewer hot pixels than actual pixels on the camera sensor. HAL must support both FAST and HIGH_QUALITY if hot pixel correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

A location object to use when generating image GPS metadata.

Setting a location object in a request will include the GPS coordinates of the location into any JPEG images captured based on the request. These coordinates can then be viewed by anyone who receives the JPEG image.

3 GPS coordinates to include in output JPEG EXIF. (-180 - 180], [-90,90], [-inf, inf] 32 characters describing GPS algorithm to include in EXIF. UTF-8 null-terminated string Time GPS fix was made to include in EXIF. UTC in seconds since January 1, 1970 The orientation for a JPEG image. Degrees in multiples of 90 0, 90, 180, 270

The clockwise rotation angle in degrees, relative to the orientation to the camera, that the JPEG picture needs to be rotated by, to be viewed upright. Camera devices may either encode this value into the JPEG EXIF header, or rotate the image data to match this orientation. When the image data is rotated, the thumbnail data will also be rotated. Note that this orientation is relative to the orientation of the camera sensor, given by android.sensor.orientation. To translate from the device orientation given by the Android sensor APIs, the following sample code may be used: private int getJpegOrientation(CameraCharacteristics c, int deviceOrientation) { if (deviceOrientation == android.view.OrientationEventListener.ORIENTATION_UNKNOWN) return 0; int sensorOrientation = c.get(CameraCharacteristics.SENSOR_ORIENTATION); // Round device orientation to a multiple of 90 deviceOrientation = (deviceOrientation + 45) / 90 * 90; // Reverse device orientation for front-facing cameras boolean facingFront = c.get(CameraCharacteristics.LENS_FACING) == CameraCharacteristics.LENS_FACING_FRONT; if (facingFront) deviceOrientation = -deviceOrientation; // Calculate desired JPEG orientation relative to camera orientation to make // the image upright relative to the device orientation int jpegOrientation = (sensorOrientation + deviceOrientation + 360) % 360; return jpegOrientation; }

Compression quality of the final JPEG image. 1-100; larger is higher quality

85-95 is typical usage range.

Compression quality of JPEG thumbnail. 1-100; larger is higher quality 2 Resolution of embedded JPEG thumbnail. android.jpeg.availableThumbnailSizes

When set to (0, 0) value, the JPEG EXIF will not contain thumbnail, but the captured JPEG will still be a valid image. For best results, when issuing a request for a JPEG image, the thumbnail size selected should have the same aspect ratio as the main JPEG output. If the thumbnail image aspect ratio differs from the JPEG primary image aspect ratio, the camera device creates the thumbnail by cropping it from the primary image. For example, if the primary image has 4:3 aspect ratio, the thumbnail image has 16:9 aspect ratio, the primary image will be cropped vertically (letterbox) to generate the thumbnail image. The thumbnail image will always have a smaller Field Of View (FOV) than the primary image when aspect ratios differ.

The HAL must not squeeze or stretch the downscaled primary image to generate thumbnail. The cropping must be done on the primary jpeg image rather than the sensor active array. The stream cropping rule specified by "S5. Cropping" in camera3.h doesn't apply to the thumbnail image cropping. 2 n List of JPEG thumbnail sizes for android.jpeg.thumbnailSize supported by this camera device.

This list will include at least one non-zero resolution, plus `(0,0)` for indicating no thumbnail should be generated. Below condiditions will be satisfied for this size list: * The sizes will be sorted by increasing pixel area (width x height). If several resolutions have the same area, they will be sorted by increasing width. * The aspect ratio of the largest thumbnail size will be same as the aspect ratio of largest JPEG output size in android.scaler.availableStreamConfigurations. The largest size is defined as the size that has the largest pixel area in a given size list. * Each output JPEG size in android.scaler.availableStreamConfigurations will have at least one corresponding size that has the same aspect ratio in availableThumbnailSizes, and vice versa. * All non-`(0, 0)` sizes will have non-zero widths and heights.

Maximum size in bytes for the compressed JPEG buffer Must be large enough to fit any JPEG produced by the camera

This is used for sizing the gralloc buffers for JPEG

The size of the compressed JPEG image, in bytes >= 0

If no JPEG output is produced for the request, this must be 0. Otherwise, this describes the real size of the compressed JPEG image placed in the output stream. More specifically, if android.jpeg.maxSize = 1000000, and a specific capture has android.jpeg.size = 500000, then the output buffer from the JPEG stream will be 1000000 bytes, of which the first 500000 make up the real data.

The desired lens aperture size, as a ratio of lens focal length to the effective aperture diameter. The f-number (f/N) android.lens.info.availableApertures

Setting this value is only supported on the camera devices that have a variable aperture lens. When this is supported and android.control.aeMode is OFF, this can be set along with android.sensor.exposureTime, android.sensor.sensitivity, and android.sensor.frameDuration to achieve manual exposure control. The requested aperture value may take several frames to reach the requested value; the camera device will report the current (intermediate) aperture size in capture result metadata while the aperture is changing. While the aperture is still changing, android.lens.state will be set to MOVING. When this is supported and android.control.aeMode is one of the ON modes, this will be overridden by the camera device auto-exposure algorithm, the overridden values are then provided back to the user in the corresponding result.

The desired setting for the lens neutral density filter(s). Exposure Value (EV) android.lens.info.availableFilterDensities

This control will not be supported on most camera devices. Lens filters are typically used to lower the amount of light the sensor is exposed to (measured in steps of EV). As used here, an EV step is the standard logarithmic representation, which are non-negative, and inversely proportional to the amount of light hitting the sensor. For example, setting this to 0 would result in no reduction of the incoming light, and setting this to 2 would mean that the filter is set to reduce incoming light by two stops (allowing 1/4 of the prior amount of light to the sensor). It may take several frames before the lens filter density changes to the requested value. While the filter density is still changing, android.lens.state will be set to MOVING.

The desired lens focal length; used for optical zoom. Millimeters android.lens.info.availableFocalLengths

This setting controls the physical focal length of the camera device's lens. Changing the focal length changes the field of view of the camera device, and is usually used for optical zoom. Like android.lens.focusDistance and android.lens.aperture, this setting won't be applied instantaneously, and it may take several frames before the lens can change to the requested focal length. While the focal length is still changing, android.lens.state will be set to MOVING. Optical zoom will not be supported on most devices.

Desired distance to plane of sharpest focus, measured from frontmost surface of the lens. See android.lens.info.focusDistanceCalibration for details >= 0

This control can be used for setting manual focus, on devices that support the MANUAL_SENSOR capability and have a variable-focus lens (see android.lens.info.minimumFocusDistance). A value of `0.0f` means infinity focus. The value set will be clamped to `[0.0f, android.lens.info.minimumFocusDistance]`. Like android.lens.focalLength, this setting won't be applied instantaneously, and it may take several frames before the lens can move to the requested focus distance. While the lens is still moving, android.lens.state will be set to MOVING. LEGACY devices support at most setting this to `0.0f` for infinity focus.

OFF Optical stabilization is unavailable. ON Optical stabilization is enabled. Sets whether the camera device uses optical image stabilization (OIS) when capturing images. android.lens.info.availableOpticalStabilization

OIS is used to compensate for motion blur due to small movements of the camera during capture. Unlike digital image stabilization (android.control.videoStabilizationMode), OIS makes use of mechanical elements to stabilize the camera sensor, and thus allows for longer exposure times before camera shake becomes apparent. Switching between different optical stabilization modes may take several frames to initialize, the camera device will report the current mode in capture result metadata. For example, When "ON" mode is requested, the optical stabilization modes in the first several capture results may still be "OFF", and it will become "ON" when the initialization is done. If a camera device supports both OIS and digital image stabilization (android.control.videoStabilizationMode), turning both modes on may produce undesirable interaction, so it is recommended not to enable both at the same time. Not all devices will support OIS; see android.lens.info.availableOpticalStabilization for available controls.

n List of aperture size values for android.lens.aperture that are supported by this camera device. The aperture f-number

If the camera device doesn't support a variable lens aperture, this list will contain only one value, which is the fixed aperture size. If the camera device supports a variable aperture, the aperture values in this list will be sorted in ascending order.

n List of neutral density filter values for android.lens.filterDensity that are supported by this camera device. Exposure value (EV) Values are >= 0

If a neutral density filter is not supported by this camera device, this list will contain only 0. Otherwise, this list will include every filter density supported by the camera device, in ascending order.

n List of focal lengths for android.lens.focalLength that are supported by this camera device. Millimeters Values are > 0

If optical zoom is not supported, this list will only contain a single value corresponding to the fixed focal length of the device. Otherwise, this list will include every focal length supported by the camera device, in ascending order.

n List of optical image stabilization (OIS) modes for android.lens.opticalStabilizationMode that are supported by this camera device. Any value listed in android.lens.opticalStabilizationMode

If OIS is not supported by a given camera device, this list will contain only OFF.

Hyperfocal distance for this lens. See android.lens.info.focusDistanceCalibration for details If lens is fixed focus, >= 0. If lens has focuser unit, the value is within `(0.0f, android.lens.info.minimumFocusDistance]`

If the lens is not fixed focus, the camera device will report this field when android.lens.info.focusDistanceCalibration is APPROXIMATE or CALIBRATED.

Shortest distance from frontmost surface of the lens that can be brought into sharp focus. See android.lens.info.focusDistanceCalibration for details >= 0

If the lens is fixed-focus, this will be 0.

Mandatory for FULL devices; LIMITED devices must always set this value to 0 for fixed-focus; and may omit the minimum focus distance otherwise. This field is also mandatory for all devices advertising the MANUAL_SENSOR capability. 2 Dimensions of lens shading map. Both values >= 1

The map should be on the order of 30-40 rows and columns, and must be smaller than 64x64.

UNCALIBRATED The lens focus distance is not accurate, and the units used for android.lens.focusDistance do not correspond to any physical units. Setting the lens to the same focus distance on separate occasions may result in a different real focus distance, depending on factors such as the orientation of the device, the age of the focusing mechanism, and the device temperature. The focus distance value will still be in the range of `[0, android.lens.info.minimumFocusDistance]`, where 0 represents the farthest focus. APPROXIMATE The lens focus distance is measured in diopters. However, setting the lens to the same focus distance on separate occasions may result in a different real focus distance, depending on factors such as the orientation of the device, the age of the focusing mechanism, and the device temperature. CALIBRATED The lens focus distance is measured in diopters, and is calibrated. The lens mechanism is calibrated so that setting the same focus distance is repeatable on multiple occasions with good accuracy, and the focus distance corresponds to the real physical distance to the plane of best focus. The lens focus distance calibration quality.

The lens focus distance calibration quality determines the reliability of focus related metadata entries, i.e. android.lens.focusDistance, android.lens.focusRange, android.lens.info.hyperfocalDistance, and android.lens.info.minimumFocusDistance. APPROXIMATE and CALIBRATED devices report the focus metadata in units of diopters (1/meter), so `0.0f` represents focusing at infinity, and increasing positive numbers represent focusing closer and closer to the camera device. The focus distance control also uses diopters on these devices. UNCALIBRATED devices do not use units that are directly comparable to any real physical measurement, but `0.0f` still represents farthest focus, and android.lens.info.minimumFocusDistance represents the nearest focus the device can achieve.

For devices advertise APPROXIMATE quality or higher, diopters 0 (infinity focus) must work. When autofocus is disabled (android.control.afMode == OFF) and the lens focus distance is set to 0 diopters (android.lens.focusDistance == 0), the lens will move to focus at infinity and is stably focused at infinity even if the device tilts. It may take the lens some time to move; during the move the lens state should be MOVING and the output diopter value should be changing toward 0. FRONT The camera device faces the same direction as the device's screen. BACK The camera device faces the opposite direction as the device's screen. EXTERNAL The camera device is an external camera, and has no fixed facing relative to the device's screen. Direction the camera faces relative to device screen. 4 The orientation of the camera relative to the sensor coordinate system. Quarternion coefficients

The four coefficients that describe the quarternion rotation from the Android sensor coordinate system to a camera-aligned coordinate system where the X-axis is aligned with the long side of the image sensor, the Y-axis is aligned with the short side of the image sensor, and the Z-axis is aligned with the optical axis of the sensor. To convert from the quarternion coefficients `(x,y,z,w)` to the axis of rotation `(a_x, a_y, a_z)` and rotation amount `theta`, the following formulas can be used: theta = 2 * acos(w) a_x = x / sin(theta/2) a_y = y / sin(theta/2) a_z = z / sin(theta/2) To create a 3x3 rotation matrix that applies the rotation defined by this quarternion, the following matrix can be used: R = [ 1 - 2y^2 - 2z^2, 2xy - 2zw, 2xz + 2yw, 2xy + 2zw, 1 - 2x^2 - 2z^2, 2yz - 2xw, 2xz - 2yw, 2yz + 2xw, 1 - 2x^2 - 2y^2 ] This matrix can then be used to apply the rotation to a column vector point with `p' = Rp` where `p` is in the device sensor coordinate system, and `p'` is in the camera-oriented coordinate system.

3 Position of the camera optical center. Meters

The position of the camera device's lens optical center, as a three-dimensional vector `(x,y,z)`, relative to the optical center of the largest camera device facing in the same direction as this camera, in the {@link android.hardware.SensorEvent Android sensor coordinate axes}. Note that only the axis definitions are shared with the sensor coordinate system, but not the origin. If this device is the largest or only camera device with a given facing, then this position will be `(0, 0, 0)`; a camera device with a lens optical center located 3 cm from the main sensor along the +X axis (to the right from the user's perspective) will report `(0.03, 0, 0)`. To transform a pixel coordinates between two cameras facing the same direction, first the source camera android.lens.radialDistortion must be corrected for. Then the source camera android.lens.intrinsicCalibration needs to be applied, followed by the android.lens.poseRotation of the source camera, the translation of the source camera relative to the destination camera, the android.lens.poseRotation of the destination camera, and finally the inverse of android.lens.intrinsicCalibration of the destination camera. This obtains a radial-distortion-free coordinate in the destination camera pixel coordinates. To compare this against a real image from the destination camera, the destination camera image then needs to be corrected for radial distortion before comparison or sampling.

Should be zero for fixed-focus cameras

2 The range of scene distances that are in sharp focus (depth of field). A pair of focus distances in diopters: (near, far); see android.lens.info.focusDistanceCalibration for details. >=0

If variable focus not supported, can still report fixed depth of field range

STATIONARY The lens parameters (android.lens.focalLength, android.lens.focusDistance, android.lens.filterDensity and android.lens.aperture) are not changing. MOVING One or several of the lens parameters (android.lens.focalLength, android.lens.focusDistance, android.lens.filterDensity or android.lens.aperture) is currently changing. Current lens status.

For lens parameters android.lens.focalLength, android.lens.focusDistance, android.lens.filterDensity and android.lens.aperture, when changes are requested, they may take several frames to reach the requested values. This state indicates the current status of the lens parameters. When the state is STATIONARY, the lens parameters are not changing. This could be either because the parameters are all fixed, or because the lens has had enough time to reach the most recently-requested values. If all these lens parameters are not changable for a camera device, as listed below: * Fixed focus (`android.lens.info.minimumFocusDistance == 0`), which means android.lens.focusDistance parameter will always be 0. * Fixed focal length (android.lens.info.availableFocalLengths contains single value), which means the optical zoom is not supported. * No ND filter (android.lens.info.availableFilterDensities contains only 0). * Fixed aperture (android.lens.info.availableApertures contains single value). Then this state will always be STATIONARY. When the state is MOVING, it indicates that at least one of the lens parameters is changing.

5 The parameters for this camera device's intrinsic calibration. Pixels in the android.sensor.info.preCorrectionActiveArraySize coordinate system.

The five calibration parameters that describe the transform from camera-centric 3D coordinates to sensor pixel coordinates: [f_x, f_y, c_x, c_y, s] Where `f_x` and `f_y` are the horizontal and vertical focal lengths, `[c_x, c_y]` is the position of the optical axis, and `s` is a skew parameter for the sensor plane not being aligned with the lens plane. These are typically used within a transformation matrix K: K = [ f_x, s, c_x, 0, f_y, c_y, 0 0, 1 ] which can then be combined with the camera pose rotation `R` and translation `t` (android.lens.poseRotation and android.lens.poseTranslation, respective) to calculate the complete transform from world coordinates to pixel coordinates: P = [ K 0 * [ R t 0 1 ] 0 1 ] and with `p_w` being a point in the world coordinate system and `p_s` being a point in the camera active pixel array coordinate system, and with the mapping including the homogeneous division by z: p_h = (x_h, y_h, z_h) = P p_w p_s = p_h / z_h so `[x_s, y_s]` is the pixel coordinates of the world point, `z_s = 1`, and `w_s` is a measurement of disparity (depth) in pixel coordinates. Note that the coordinate system for this transform is the android.sensor.info.preCorrectionActiveArraySize system, where `(0,0)` is the top-left of the preCorrectionActiveArraySize rectangle. Once the pose and intrinsic calibration transforms have been applied to a world point, then the android.lens.radialDistortion transform needs to be applied, and the result adjusted to be in the android.sensor.info.activeArraySize coordinate system (where `(0, 0)` is the top-left of the activeArraySize rectangle), to determine the final pixel coordinate of the world point for processed (non-RAW) output buffers.

6 The correction coefficients to correct for this camera device's radial and tangential lens distortion. Unitless coefficients.

Four radial distortion coefficients `[kappa_0, kappa_1, kappa_2, kappa_3]` and two tangential distortion coefficients `[kappa_4, kappa_5]` that can be used to correct the lens's geometric distortion with the mapping equations: x_c = x_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) + kappa_4 * (2 * x_i * y_i) + kappa_5 * ( r^2 + 2 * x_i^2 ) y_c = y_i * ( kappa_0 + kappa_1 * r^2 + kappa_2 * r^4 + kappa_3 * r^6 ) + kappa_5 * (2 * x_i * y_i) + kappa_4 * ( r^2 + 2 * y_i^2 ) Here, `[x_c, y_c]` are the coordinates to sample in the input image that correspond to the pixel values in the corrected image at the coordinate `[x_i, y_i]`: correctedImage(x_i, y_i) = sample_at(x_c, y_c, inputImage) The pixel coordinates are defined in a normalized coordinate system related to the android.lens.intrinsicCalibration calibration fields. Both `[x_i, y_i]` and `[x_c, y_c]` have `(0,0)` at the lens optical center `[c_x, c_y]`. The maximum magnitudes of both x and y coordinates are normalized to be 1 at the edge further from the optical center, so the range for both dimensions is `-1 <= x <= 1`. Finally, `r` represents the radial distance from the optical center, `r^2 = x_i^2 + y_i^2`, and its magnitude is therefore no larger than `|r| <= sqrt(2)`. The distortion model used is the Brown-Conrady model.

OFF No noise reduction is applied. FAST Noise reduction is applied without reducing frame rate relative to sensor output. HIGH_QUALITY High-quality noise reduction is applied, at the cost of possibly reduced frame rate relative to sensor output. MINIMAL MINIMAL noise reduction is applied without reducing frame rate relative to sensor output. Mode of operation for the noise reduction algorithm. android.noiseReduction.availableNoiseReductionModes

The noise reduction algorithm attempts to improve image quality by removing excessive noise added by the capture process, especially in dark conditions. OFF means no noise reduction will be applied by the camera device, for both raw and YUV domain. MINIMAL means that only sensor raw domain basic noise reduction is enabled ,to remove demosaicing or other processing artifacts. For YUV_REPROCESSING, MINIMAL is same as OFF. This mode is optional, may not be support by all devices. The application should check android.noiseReduction.availableNoiseReductionModes before using it. FAST/HIGH_QUALITY both mean camera device determined noise filtering will be applied. HIGH_QUALITY mode indicates that the camera device will use the highest-quality noise filtering algorithms, even if it slows down capture rate. FAST means the camera device will not slow down capture rate when applying noise filtering. For YUV_REPROCESSING, these FAST/HIGH_QUALITY modes both mean that the camera device will apply FAST/HIGH_QUALITY YUV domain noise reduction, respectively. The camera device may adjust the noise reduction parameters for best image quality based on the android.reprocess.effectiveExposureFactor if it is set.

For YUV_REPROCESSING The HAL can use android.reprocess.effectiveExposureFactor to adjust the internal noise reduction parameters appropriately to get the best quality images. Control the amount of noise reduction applied to the images 1-10; 10 is max noise reduction 1 - 10 n List of noise reduction modes for android.noiseReduction.mode that are supported by this camera device. Any value listed in android.noiseReduction.mode

Full-capability camera devices will always support OFF and FAST. Legacy-capability camera devices will only support FAST mode.

HAL must support both FAST and HIGH_QUALITY if noise reduction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

If set to 1, the camera service does not scale 'normalized' coordinates with respect to the crop region. This applies to metering input (a{e,f,wb}Region and output (face rectangles).

Normalized coordinates refer to those in the (-1000,1000) range mentioned in the android.hardware.Camera API. HAL implementations should instead always use and emit sensor array-relative coordinates for all region data. Does not need to be listed in static metadata. Support will be removed in future versions of camera service.

If set to 1, then the camera service always switches to FOCUS_MODE_AUTO before issuing a AF trigger.

HAL implementations should implement AF trigger modes for AUTO, MACRO, CONTINUOUS_FOCUS, and CONTINUOUS_PICTURE modes instead of using this flag. Does not need to be listed in static metadata. Support will be removed in future versions of camera service

If set to 1, the camera service uses CAMERA2_PIXEL_FORMAT_ZSL instead of HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED for the zero shutter lag stream

HAL implementations should use gralloc usage flags to determine that a stream will be used for zero-shutter-lag, instead of relying on an explicit format setting. Does not need to be listed in static metadata. Support will be removed in future versions of camera service.

If set to 1, the HAL will always split result metadata for a single capture into multiple buffers, returned using multiple process_capture_result calls.

Does not need to be listed in static metadata. Support for partial results will be reworked in future versions of camera service. This quirk will stop working at that point; DO NOT USE without careful consideration of future support.

Refer to `camera3_capture_result::partial_result` for information on how to implement partial results. FINAL The last or only metadata result buffer for this capture. PARTIAL A partial buffer of result metadata for this capture. More result buffers for this capture will be sent by the camera device, the last of which will be marked FINAL. Whether a result given to the framework is the final one for the capture, or only a partial that contains a subset of the full set of dynamic metadata values. Optional. Default value is FINAL.

The entries in the result metadata buffers for a single capture may not overlap, except for this entry. The FINAL buffers must retain FIFO ordering relative to the requests that generate them, so the FINAL buffer for frame 3 must always be sent to the framework after the FINAL buffer for frame 2, and before the FINAL buffer for frame 4. PARTIAL buffers may be returned in any order relative to other frames, but all PARTIAL buffers for a given capture must arrive before the FINAL buffer for that capture. This entry may only be used by the camera device if quirks.usePartialResult is set to 1.

Refer to `camera3_capture_result::partial_result` for information on how to implement partial results.

A frame counter set by the framework. Must be maintained unchanged in output frame. This value monotonically increases with every new result (that is, each new result has a unique frameCount value). incrementing integer Any int. An application-specified ID for the current request. Must be maintained unchanged in output frame arbitrary integer assigned by application Any int n List which camera reprocess stream is used for the source of reprocessing data. List of camera reprocess stream IDs Typically, only one entry allowed, must be a valid reprocess stream ID.

Only meaningful when android.request.type == REPROCESS. Ignored otherwise

NONE No metadata should be produced on output, except for application-bound buffer data. If no application-bound streams exist, no frame should be placed in the output frame queue. If such streams exist, a frame should be placed on the output queue with null metadata but with the necessary output buffer information. Timestamp information should still be included with any output stream buffers FULL All metadata should be produced. Statistics will only be produced if they are separately enabled How much metadata to produce on output n Lists which camera output streams image data from this capture must be sent to List of camera stream IDs List must only include streams that have been created

If no output streams are listed, then the image data should simply be discarded. The image data must still be captured for metadata and statistics production, and the lens and flash must operate as requested.

CAPTURE Capture a new image from the imaging hardware, and process it according to the settings REPROCESS Process previously captured data; the android.request.inputStreams parameter determines the source reprocessing stream. TODO: Mark dynamic metadata needed for reprocessing with [RP] The type of the request; either CAPTURE or REPROCESS. For HAL3, this tag is redundant. 3 The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device. For processed (and stalling) format streams, >= 1. For Raw format (either stalling or non-stalling) streams, >= 0. For processed (but not stalling) format streams, >= 3 for FULL mode devices (`android.info.supportedHardwareLevel == FULL`); >= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).

This is a 3 element tuple that contains the max number of output simultaneous streams for raw sensor, processed (but not stalling), and processed (and stalling) formats respectively. For example, assuming that JPEG is typically a processed and stalling stream, if max raw sensor format output stream number is 1, max YUV streams number is 3, and max JPEG stream number is 2, then this tuple should be `(1, 3, 2)`. This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for an output stream can be any supported format provided by android.scaler.availableStreamConfigurations. The formats defined in android.scaler.availableStreamConfigurations can be catergorized into the 3 stream types as below: * Processed (but stalling): any non-RAW format with a stallDurations > 0. Typically {@link android.graphics.ImageFormat#JPEG JPEG format}. * Raw formats: {@link android.graphics.ImageFormat#RAW_SENSOR RAW_SENSOR}, {@link android.graphics.ImageFormat#RAW10 RAW10}, or {@link android.graphics.ImageFormat#RAW12 RAW12}. * Processed (but not-stalling): any non-RAW format without a stall duration. Typically {@link android.graphics.ImageFormat#YUV_420_888 YUV_420_888}, {@link android.graphics.ImageFormat#NV21 NV21}, or {@link android.graphics.ImageFormat#YV12 YV12}.

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any `RAW` formats. >= 0

This value contains the max number of output simultaneous streams from the raw sensor. This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any `RAW` and supported format provided by android.scaler.streamConfigurationMap. In particular, a `RAW` format is typically one of: * {@link android.graphics.ImageFormat#RAW_SENSOR RAW_SENSOR} * {@link android.graphics.ImageFormat#RAW10 RAW10} * {@link android.graphics.ImageFormat#RAW12 RAW12} LEGACY mode devices (android.info.supportedHardwareLevel `==` LEGACY) never support raw streams.

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any processed (but not-stalling) formats. >= 3 for FULL mode devices (`android.info.supportedHardwareLevel == FULL`); >= 2 for LIMITED mode devices (`android.info.supportedHardwareLevel == LIMITED`).

This value contains the max number of output simultaneous streams for any processed (but not-stalling) formats. This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap. Processed (but not-stalling) is defined as any non-RAW format without a stall duration. Typically: * {@link android.graphics.ImageFormat#YUV_420_888 YUV_420_888} * {@link android.graphics.ImageFormat#NV21 NV21} * {@link android.graphics.ImageFormat#YV12 YV12} * Implementation-defined formats, i.e. {@link android.hardware.camera2.params.StreamConfigurationMap#isOutputSupportedFor(Class)} For full guarantees, query {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} with a processed format -- it will return 0 for a non-stalling stream. LEGACY devices will support at least 2 processing/non-stalling streams.

The maximum numbers of different types of output streams that can be configured and used simultaneously by a camera device for any processed (and stalling) formats. >= 1

This value contains the max number of output simultaneous streams for any processed (but not-stalling) formats. This lists the upper bound of the number of output streams supported by the camera device. Using more streams simultaneously may require more hardware and CPU resources that will consume more power. The image format for this kind of an output stream can be any non-`RAW` and supported format provided by android.scaler.streamConfigurationMap. A processed and stalling format is defined as any non-RAW format with a stallDurations > 0. Typically only the {@link android.graphics.ImageFormat#JPEG JPEG format} is a stalling format. For full guarantees, query {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} with a processed format -- it will return a non-0 value for a stalling stream. LEGACY devices will support up to 1 processing/stalling stream.

1 How many reprocessing streams of any type can be allocated at the same time. >= 0

Only used by HAL2.x. When set to 0, it means no reprocess stream is supported.

The maximum numbers of any type of input streams that can be configured and used simultaneously by a camera device. 0 or 1.

When set to 0, it means no input stream is supported. The image format for a input stream can be any supported format returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}. When using an input stream, there must be at least one output stream configured to to receive the reprocessed images. When an input stream and some output streams are used in a reprocessing request, only the input buffer will be used to produce these output stream buffers, and a new sensor image will not be captured. For example, for Zero Shutter Lag (ZSL) still capture use case, the input stream image format will be PRIVATE, the associated output stream image format should be JPEG.

For the reprocessing flow and controls, see hardware/libhardware/include/hardware/camera3.h Section 10 for more details. A frame counter set by the framework. This value monotonically increases with every new result (that is, each new result has a unique frameCount value). count of frames > 0

Reset on release()

Specifies the number of pipeline stages the frame went through from when it was exposed to when the final completed result was available to the framework. <= android.request.pipelineMaxDepth

Depending on what settings are used in the request, and what streams are configured, the data may undergo less processing, and some pipeline stages skipped. See android.request.pipelineMaxDepth for more details.

This value must always represent the accurate count of how many pipeline stages were actually used. Specifies the number of maximum pipeline stages a frame has to go through from when it's exposed to when it's available to the framework.

A typical minimum value for this is 2 (one stage to expose, one stage to readout) from the sensor. The ISP then usually adds its own stages to do custom HW processing. Further stages may be added by SW processing. Depending on what settings are used (e.g. YUV, JPEG) and what processing is enabled (e.g. face detection), the actual pipeline depth (specified by android.request.pipelineDepth) may be less than the max pipeline depth. A pipeline depth of X stages is equivalent to a pipeline latency of X frame intervals. This value will normally be 8 or less, however, for high speed capture session, the max pipeline depth will be up to 8 x size of high speed capture request list.

This value should be 4 or less, expect for the high speed recording session, where the max batch sizes may be larger than 1. Defines how many sub-components a result will be composed of. >= 1

In order to combat the pipeline latency, partial results may be delivered to the application layer from the camera device as soon as they are available. Optional; defaults to 1. A value of 1 means that partial results are not supported, and only the final TotalCaptureResult will be produced by the camera device. A typical use case for this might be: after requesting an auto-focus (AF) lock the new AF state might be available 50% of the way through the pipeline. The camera device could then immediately dispatch this state via a partial result to the application, and the rest of the metadata via later partial results.

n BACKWARD_COMPATIBLE The minimal set of capabilities that every camera device (regardless of android.info.supportedHardwareLevel) supports. This capability is listed by all normal devices, and indicates that the camera device has a feature set that's comparable to the baseline requirements for the older android.hardware.Camera API. Devices with the DEPTH_OUTPUT capability might not list this capability, indicating that they support only depth measurement, not standard color output. MANUAL_SENSOR The camera device can be manually controlled (3A algorithms such as auto-exposure, and auto-focus can be bypassed). The camera device supports basic manual control of the sensor image acquisition related stages. This means the following controls are guaranteed to be supported: * Manual frame duration control * android.sensor.frameDuration * android.sensor.info.maxFrameDuration * Manual exposure control * android.sensor.exposureTime * android.sensor.info.exposureTimeRange * Manual sensitivity control * android.sensor.sensitivity * android.sensor.info.sensitivityRange * Manual lens control (if the lens is adjustable) * android.lens.* * Manual flash control (if a flash unit is present) * android.flash.* * Manual black level locking * android.blackLevel.lock * Auto exposure lock * android.control.aeLock If any of the above 3A algorithms are enabled, then the camera device will accurately report the values applied by 3A in the result. A given camera device may also support additional manual sensor controls, but this capability only covers the above list of controls. If this is supported, android.scaler.streamConfigurationMap will additionally return a min frame duration that is greater than zero for each supported size-format combination. MANUAL_POST_PROCESSING The camera device post-processing stages can be manually controlled. The camera device supports basic manual control of the image post-processing stages. This means the following controls are guaranteed to be supported: * Manual tonemap control * android.tonemap.curve * android.tonemap.mode * android.tonemap.maxCurvePoints * android.tonemap.gamma * android.tonemap.presetCurve * Manual white balance control * android.colorCorrection.transform * android.colorCorrection.gains * Manual lens shading map control * android.shading.mode * android.statistics.lensShadingMapMode * android.statistics.lensShadingMap * android.lens.info.shadingMapSize * Manual aberration correction control (if aberration correction is supported) * android.colorCorrection.aberrationMode * android.colorCorrection.availableAberrationModes * Auto white balance lock * android.control.awbLock If auto white balance is enabled, then the camera device will accurately report the values applied by AWB in the result. A given camera device may also support additional post-processing controls, but this capability only covers the above list of controls. RAW The camera device supports outputting RAW buffers and metadata for interpreting them. Devices supporting the RAW capability allow both for saving DNG files, and for direct application processing of raw sensor images. * RAW_SENSOR is supported as an output format. * The maximum available resolution for RAW_SENSOR streams will match either the value in android.sensor.info.pixelArraySize or android.sensor.info.activeArraySize. * All DNG-related optional metadata entries are provided by the camera device. PRIVATE_REPROCESSING The camera device supports the Zero Shutter Lag reprocessing use case. * One input stream is supported, that is, `android.request.maxNumInputStreams == 1`. * {@link android.graphics.ImageFormat#PRIVATE} is supported as an output/input format, that is, {@link android.graphics.ImageFormat#PRIVATE} is included in the lists of formats returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputFormats} and {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputFormats}. * {@link android.hardware.camera2.params.StreamConfigurationMap#getValidOutputFormatsForInput} returns non empty int[] for each supported input format returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}. * Each size returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputSizes getInputSizes(ImageFormat.PRIVATE)} is also included in {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes getOutputSizes(ImageFormat.PRIVATE)} * Using {@link android.graphics.ImageFormat#PRIVATE} does not cause a frame rate drop relative to the sensor's maximum capture rate (at that resolution). * {@link android.graphics.ImageFormat#PRIVATE} will be reprocessable into both {@link android.graphics.ImageFormat#YUV_420_888} and {@link android.graphics.ImageFormat#JPEG} formats. * The maximum available resolution for PRIVATE streams (both input/output) will match the maximum available resolution of JPEG streams. * Static metadata android.reprocess.maxCaptureStall. * Only below controls are effective for reprocessing requests and will be present in capture results, other controls in reprocess requests will be ignored by the camera device. * android.jpeg.* * android.noiseReduction.mode * android.edge.mode READ_SENSOR_SETTINGS The camera device supports accurately reporting the sensor settings for many of the sensor controls while the built-in 3A algorithm is running. This allows reporting of sensor settings even when these settings cannot be manually changed. The values reported for the following controls are guaranteed to be available in the CaptureResult, including when 3A is enabled: * Exposure control * android.sensor.exposureTime * Sensitivity control * android.sensor.sensitivity * Lens controls (if the lens is adjustable) * android.lens.focusDistance * android.lens.aperture This capability is a subset of the MANUAL_SENSOR control capability, and will always be included if the MANUAL_SENSOR capability is available. BURST_CAPTURE The camera device supports capturing maximum-resolution images at >= 20 frames per second, in at least the uncompressed YUV format, when post-processing settings are set to FAST. More specifically, this means that a size matching the camera device's active array size is listed as a supported size for the YUV_420_888 format in android.scaler.streamConfigurationMap, the minimum frame duration for that format and size is <= 1/20 s, and the android.control.aeAvailableTargetFpsRanges entry lists at least one FPS range where the minimum FPS is >= 1 / minimumFrameDuration for the maximum-size YUV_420_888 format. In addition, the android.sync.maxLatency field is guaranted to have a value between 0 and 4, inclusive. android.control.aeLockAvailable and android.control.awbLockAvailable are also guaranteed to be `true` so burst capture with these two locks ON yields consistent image output. On a camera device that reports the HIGH_RESOLUTION hardware level, meaning the device supports very large capture sizes, BURST_CAPTURE means that at least 8-megapixel images can be captured at `>=` 20 fps, and maximum-resolution images can be captured at `>=` 10 fps. YUV_REPROCESSING The camera device supports the YUV_420_888 reprocessing use case, similar as PRIVATE_REPROCESSING, This capability requires the camera device to support the following: * One input stream is supported, that is, `android.request.maxNumInputStreams == 1`. * {@link android.graphics.ImageFormat#YUV_420_888} is supported as an output/input format, that is, YUV_420_888 is included in the lists of formats returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputFormats} and {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputFormats}. * {@link android.hardware.camera2.params.StreamConfigurationMap#getValidOutputFormatsForInput} returns non-empty int[] for each supported input format returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputFormats}. * Each size returned by {@link android.hardware.camera2.params.StreamConfigurationMap#getInputSizes getInputSizes(YUV_420_888)} is also included in {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputSizes getOutputSizes(YUV_420_888)} * Using {@link android.graphics.ImageFormat#YUV_420_888} does not cause a frame rate drop relative to the sensor's maximum capture rate (at that resolution). * {@link android.graphics.ImageFormat#YUV_420_888} will be reprocessable into both {@link android.graphics.ImageFormat#YUV_420_888} and {@link android.graphics.ImageFormat#JPEG} formats. * The maximum available resolution for {@link android.graphics.ImageFormat#YUV_420_888} streams (both input/output) will match the maximum available resolution of {@link android.graphics.ImageFormat#JPEG} streams. * Static metadata android.reprocess.maxCaptureStall. * Only the below controls are effective for reprocessing requests and will be present in capture results. The reprocess requests are from the original capture results that are associated with the intermediate {@link android.graphics.ImageFormat#YUV_420_888} output buffers. All other controls in the reprocess requests will be ignored by the camera device. * android.jpeg.* * android.noiseReduction.mode * android.edge.mode * android.reprocess.effectiveExposureFactor DEPTH_OUTPUT The camera device can produce depth measurements from its field of view. This capability requires the camera device to support the following: * {@link android.graphics.ImageFormat#DEPTH16} is supported as an output format. * {@link android.graphics.ImageFormat#DEPTH_POINT_CLOUD} is optionally supported as an output format. * This camera device, and all camera devices with the same android.lens.facing, will list the following calibration entries in both {@link android.hardware.camera2.CameraCharacteristics} and {@link android.hardware.camera2.CaptureResult}: - android.lens.poseTranslation - android.lens.poseRotation - android.lens.intrinsicCalibration - android.lens.radialDistortion * The android.depth.depthIsExclusive entry is listed by this device. * A LIMITED camera with only the DEPTH_OUTPUT capability does not have to support normal YUV_420_888, JPEG, and PRIV-format outputs. It only has to support the DEPTH16 format. Generally, depth output operates at a slower frame rate than standard color capture, so the DEPTH16 and DEPTH_POINT_CLOUD formats will commonly have a stall duration that should be accounted for (see {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration}). On a device that supports both depth and color-based output, to enable smooth preview, using a repeating burst is recommended, where a depth-output target is only included once every N frames, where N is the ratio between preview output rate and depth output rate, including depth stall time. CONSTRAINED_HIGH_SPEED_VIDEO The device supports constrained high speed video recording (frame rate >=120fps) use case. The camera device will support high speed capture session created by {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}, which only accepts high speed request list created by {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedRequestList}. A camera device can still support high speed video streaming by advertising the high speed FPS ranges in android.control.aeAvailableTargetFpsRanges. For this case, all normal capture request per frame control and synchronization requirements will apply to the high speed fps ranges, the same as all other fps ranges. This capability describes the capability of a specialized operating mode with many limitations (see below), which is only targeted at high speed video recording. The supported high speed video sizes and fps ranges are specified in {@link android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoFpsRanges}. To get desired output frame rates, the application is only allowed to select video size and FPS range combinations provided by {@link android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoSizes}. The fps range can be controlled via android.control.aeTargetFpsRange. In this capability, the camera device will override aeMode, awbMode, and afMode to ON, ON, and CONTINUOUS_VIDEO, respectively. All post-processing block mode controls will be overridden to be FAST. Therefore, no manual control of capture and post-processing parameters is possible. All other controls operate the same as when android.control.mode == AUTO. This means that all other android.control.* fields continue to work, such as * android.control.aeTargetFpsRange * android.control.aeExposureCompensation * android.control.aeLock * android.control.awbLock * android.control.effectMode * android.control.aeRegions * android.control.afRegions * android.control.awbRegions * android.control.afTrigger * android.control.aePrecaptureTrigger Outside of android.control.*, the following controls will work: * android.flash.mode (TORCH mode only, automatic flash for still capture will not work since aeMode is ON) * android.lens.opticalStabilizationMode (if it is supported) * android.scaler.cropRegion * android.statistics.faceDetectMode (if it is supported) For high speed recording use case, the actual maximum supported frame rate may be lower than what camera can output, depending on the destination Surfaces for the image data. For example, if the destination surface is from video encoder, the application need check if the video encoder is capable of supporting the high frame rate for a given video size, or it will end up with lower recording frame rate. If the destination surface is from preview window, the actual preview frame rate will be bounded by the screen refresh rate. The camera device will only support up to 2 high speed simultaneous output surfaces (preview and recording surfaces) in this mode. Above controls will be effective only if all of below conditions are true: * The application creates a camera capture session with no more than 2 surfaces via {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession}. The targeted surfaces must be preview surface (either from {@link android.view.SurfaceView} or {@link android.graphics.SurfaceTexture}) or recording surface(either from {@link android.media.MediaRecorder#getSurface} or {@link android.media.MediaCodec#createInputSurface}). * The stream sizes are selected from the sizes reported by {@link android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoSizes}. * The FPS ranges are selected from {@link android.hardware.camera2.params.StreamConfigurationMap#getHighSpeedVideoFpsRanges}. When above conditions are NOT satistied, the {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedCaptureSession} and {@link android.hardware.camera2.CameraDevice#createConstrainedHighSpeedRequestList} will fail. Switching to a FPS range that has different maximum FPS may trigger some camera device reconfigurations, which may introduce extra latency. It is recommended that the application avoids unnecessary maximum target FPS changes as much as possible during high speed streaming. List of capabilities that this camera device advertises as fully supporting.

A capability is a contract that the camera device makes in order to be able to satisfy one or more use cases. Listing a capability guarantees that the whole set of features required to support a common use will all be available. Using a subset of the functionality provided by an unsupported capability may be possible on a specific camera device implementation; to do this query each of android.request.availableRequestKeys, android.request.availableResultKeys, android.request.availableCharacteristicsKeys. The following capabilities are guaranteed to be available on android.info.supportedHardwareLevel `==` FULL devices: * MANUAL_SENSOR * MANUAL_POST_PROCESSING Other capabilities may be available on either FULL or LIMITED devices, but the application should query this key to be sure.

Additional constraint details per-capability will be available in the Compatibility Test Suite. Minimum baseline requirements required for the BACKWARD_COMPATIBLE capability are not explicitly listed. Instead refer to "BC" tags and the camera CTS tests in the android.hardware.camera2.cts package. Listed controls that can be either request or result (e.g. android.sensor.exposureTime) must be available both in the request and the result in order to be considered to be capability-compliant. For example, if the HAL claims to support MANUAL control, then exposure time must be configurable via the request _and_ the actual exposure applied must be available via the result. If MANUAL_SENSOR is omitted, the HAL may choose to omit the android.scaler.availableMinFrameDurations static property entirely. For PRIVATE_REPROCESSING and YUV_REPROCESSING capabilities, see hardware/libhardware/include/hardware/camera3.h Section 10 for more information. Devices that support the MANUAL_SENSOR capability must support the CAMERA3_TEMPLATE_MANUAL template defined in camera3.h. Devices that support the PRIVATE_REPROCESSING capability or the YUV_REPROCESSING capability must support the CAMERA3_TEMPLATE_ZERO_SHUTTER_LAG template defined in camera3.h. For DEPTH_OUTPUT, the depth-format keys android.depth.availableDepthStreamConfigurations, android.depth.availableDepthMinFrameDurations, android.depth.availableDepthStallDurations must be available, in addition to the other keys explicitly mentioned in the DEPTH_OUTPUT enum notes. The entry android.depth.maxDepthSamples must be available if the DEPTH_POINT_CLOUD format is supported (HAL pixel format BLOB, dataspace DEPTH). n A list of all keys that the camera device has available to use with {@link android.hardware.camera2.CaptureRequest}.

Attempting to set a key into a CaptureRequest that is not listed here will result in an invalid request and will be rejected by the camera device. This field can be used to query the feature set of a camera device at a more granular level than capabilities. This is especially important for optional keys that are not listed under any capability in android.request.availableCapabilities.

Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details). Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field. The HAL must not consume any request tags that are not listed either here or in the vendor tag list. The public camera2 API will always make the vendor tags visible via {@link android.hardware.camera2.CameraCharacteristics#getAvailableCaptureRequestKeys}. n A list of all keys that the camera device has available to use with {@link android.hardware.camera2.CaptureResult}.

Attempting to get a key from a CaptureResult that is not listed here will always return a `null` value. Getting a key from a CaptureResult that is listed here will generally never return a `null` value. The following keys may return `null` unless they are enabled: * android.statistics.lensShadingMap (non-null iff android.statistics.lensShadingMapMode == ON) (Those sometimes-null keys will nevertheless be listed here if they are available.) This field can be used to query the feature set of a camera device at a more granular level than capabilities. This is especially important for optional keys that are not listed under any capability in android.request.availableCapabilities.

Tags listed here must always have an entry in the result metadata, even if that size is 0 elements. Only array-type tags (e.g. lists, matrices, strings) are allowed to have 0 elements. Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details). Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field. The HAL must not produce any result tags that are not listed either here or in the vendor tag list. The public camera2 API will always make the vendor tags visible via {@link android.hardware.camera2.CameraCharacteristics#getAvailableCaptureResultKeys}. n A list of all keys that the camera device has available to use with {@link android.hardware.camera2.CameraCharacteristics}.

This entry follows the same rules as android.request.availableResultKeys (except that it applies for CameraCharacteristics instead of CaptureResult). See above for more details.

Keys listed here must always have an entry in the static info metadata, even if that size is 0 elements. Only array-type tags (e.g. lists, matrices, strings) are allowed to have 0 elements. Vendor tags must not be listed here. Use the vendor tag metadata extensions C api instead (refer to camera3.h for more details). Setting/getting vendor tags will be checked against the metadata vendor extensions API and not against this field. The HAL must not have any tags in its static info that are not listed either here or in the vendor tag list. The public camera2 API will always make the vendor tags visible via {@link android.hardware.camera2.CameraCharacteristics#getKeys}.

4 The desired region of the sensor to read out for this capture. Pixel coordinates relative to android.sensor.info.activeArraySize

This control can be used to implement digital zoom. The crop region coordinate system is based off android.sensor.info.activeArraySize, with `(0, 0)` being the top-left corner of the sensor active array. Output streams use this rectangle to produce their output, cropping to a smaller region if necessary to maintain the stream's aspect ratio, then scaling the sensor input to match the output's configured resolution. The crop region is applied after the RAW to other color space (e.g. YUV) conversion. Since raw streams (e.g. RAW16) don't have the conversion stage, they are not croppable. The crop region will be ignored by raw streams. For non-raw streams, any additional per-stream cropping will be done to maximize the final pixel area of the stream. For example, if the crop region is set to a 4:3 aspect ratio, then 4:3 streams will use the exact crop region. 16:9 streams will further crop vertically (letterbox). Conversely, if the crop region is set to a 16:9, then 4:3 outputs will crop horizontally (pillarbox), and 16:9 streams will match exactly. These additional crops will be centered within the crop region. The width and height of the crop region cannot be set to be smaller than `floor( activeArraySize.width / android.scaler.availableMaxDigitalZoom )` and `floor( activeArraySize.height / android.scaler.availableMaxDigitalZoom )`, respectively. The camera device may adjust the crop region to account for rounding and other hardware requirements; the final crop region used will be included in the output capture result.

The output streams must maintain square pixels at all times, no matter what the relative aspect ratios of the crop region and the stream are. Negative values for corner are allowed for raw output if full pixel array is larger than active pixel array. Width and height may be rounded to nearest larger supportable width, especially for raw output, where only a few fixed scales may be possible. For a set of output streams configured, if the sensor output is cropped to a smaller size than active array size, the HAL need follow below cropping rules: * The HAL need handle the cropRegion as if the sensor crop size is the effective active array size.More specifically, the HAL must transform the request cropRegion from android.sensor.info.activeArraySize to the sensor cropped pixel area size in this way: 1. Translate the requested cropRegion w.r.t., the left top corner of the sensor cropped pixel area by (tx, ty), where `tx = sensorCrop.top * (sensorCrop.height / activeArraySize.height)` and `tx = sensorCrop.left * (sensorCrop.width / activeArraySize.width)`. The (sensorCrop.top, sensorCrop.left) is the coordinate based off the android.sensor.info.activeArraySize. 2. Scale the width and height of requested cropRegion with scaling factor of sensorCrop.width/activeArraySize.width and sensorCrop.height/activeArraySize.height respectively. Once this new cropRegion is calculated, the HAL must use this region to crop the image with regard to the sensor crop size (effective active array size). The HAL still need follow the general cropping rule for this new cropRegion and effective active array size. * The HAL must report the cropRegion with regard to android.sensor.info.activeArraySize. The HAL need convert the new cropRegion generated above w.r.t., full active array size. The reported cropRegion may be slightly different with the requested cropRegion since the HAL may adjust the crop region to account for rounding, conversion error, or other hardware limitations. HAL2.x uses only (x, y, width) n RAW16 RAW16 is a standard, cross-platform format for raw image buffers with 16-bit pixels. Buffers of this format are typically expected to have a Bayer Color Filter Array (CFA) layout, which is given in android.sensor.info.colorFilterArrangement. Sensors with CFAs that are not representable by a format in android.sensor.info.colorFilterArrangement should not use this format. Buffers of this format will also follow the constraints given for RAW_OPAQUE buffers, but with relaxed performance constraints. This format is intended to give users access to the full contents of the buffers coming directly from the image sensor prior to any cropping or scaling operations, and all coordinate systems for metadata used for this format are relative to the size of the active region of the image sensor before any geometric distortion correction has been applied (i.e. android.sensor.info.preCorrectionActiveArraySize). Supported dimensions for this format are limited to the full dimensions of the sensor (e.g. either android.sensor.info.pixelArraySize or android.sensor.info.preCorrectionActiveArraySize will be the only supported output size). See android.scaler.availableInputOutputFormatsMap for the full set of performance guarantees. RAW_OPAQUE RAW_OPAQUE is a format for raw image buffers coming from an image sensor. The actual structure of buffers of this format is platform-specific, but must follow several constraints: 1. No image post-processing operations may have been applied to buffers of this type. These buffers contain raw image data coming directly from the image sensor. 1. If a buffer of this format is passed to the camera device for reprocessing, the resulting images will be identical to the images produced if the buffer had come directly from the sensor and was processed with the same settings. The intended use for this format is to allow access to the native raw format buffers coming directly from the camera sensor without any additional conversions or decrease in framerate. See android.scaler.availableInputOutputFormatsMap for the full set of performance guarantees. YV12 YCrCb 4:2:0 Planar YCrCb_420_SP NV21 IMPLEMENTATION_DEFINED System internal format, not application-accessible YCbCr_420_888 Flexible YUV420 Format BLOB JPEG format The list of image formats that are supported by this camera device for output streams.

All camera devices will support JPEG and YUV_420_888 formats. When set to YUV_420_888, application can access the YUV420 data directly.

These format values are from HAL_PIXEL_FORMAT_* in system/core/include/system/graphics.h. When IMPLEMENTATION_DEFINED is used, the platform gralloc module will select a format based on the usage flags provided by the camera HAL device and the other endpoint of the stream. It is usually used by preview and recording streams, where the application doesn't need access the image data. YCbCr_420_888 format must be supported by the HAL. When an image stream needs CPU/application direct access, this format will be used. The BLOB format must be supported by the HAL. This is used for the JPEG stream. A RAW_OPAQUE buffer should contain only pixel data. It is strongly recommended that any information used by the camera device when processing images is fully expressed by the result metadata for that image buffer. n The minimum frame duration that is supported for each resolution in android.scaler.availableJpegSizes. Nanoseconds TODO: Remove property.

This corresponds to the minimum steady-state frame duration when only that JPEG stream is active and captured in a burst, with all processing (typically in android.*.mode) set to FAST. When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations)

n 2 The JPEG resolutions that are supported by this camera device. TODO: Remove property.

The resolutions are listed as `(width, height)` pairs. All camera devices will support sensor maximum resolution (defined by android.sensor.info.activeArraySize).

The HAL must include sensor maximum resolution (defined by android.sensor.info.activeArraySize), and should include half/quarter of sensor maximum resolution. The maximum ratio between both active area width and crop region width, and active area height and crop region height, for android.scaler.cropRegion. Zoom scale factor >=1

This represents the maximum amount of zooming possible by the camera device, or equivalently, the minimum cropping window size. Crop regions that have a width or height that is smaller than this ratio allows will be rounded up to the minimum allowed size by the camera device.

n For each available processed output size (defined in android.scaler.availableProcessedSizes), this property lists the minimum supportable frame duration for that size. Nanoseconds

This should correspond to the frame duration when only that processed stream is active, with all processing (typically in android.*.mode) set to FAST. When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations).

n 2 The resolutions available for use with processed output streams, such as YV12, NV12, and platform opaque YUV/RGB streams to the GPU or video encoders.

The resolutions are listed as `(width, height)` pairs. For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide. Please reference the documentation for the image data destination to check if it limits the maximum size for image data.

For FULL capability devices (`android.info.supportedHardwareLevel == FULL`), the HAL must include all JPEG sizes listed in android.scaler.availableJpegSizes and each below resolution if it is smaller than or equal to the sensor maximum resolution (if they are not listed in JPEG sizes already): * 240p (320 x 240) * 480p (640 x 480) * 720p (1280 x 720) * 1080p (1920 x 1080) For LIMITED capability devices (`android.info.supportedHardwareLevel == LIMITED`), the HAL only has to list up to the maximum video size supported by the devices. n For each available raw output size (defined in android.scaler.availableRawSizes), this property lists the minimum supportable frame duration for that size. Nanoseconds

Should correspond to the frame duration when only the raw stream is active. When multiple streams are configured, the minimum frame duration will be >= max(individual stream min durations)

n 2 The resolutions available for use with raw sensor output streams, listed as width, height The mapping of image formats that are supported by this camera device for input streams, to their corresponding output formats.

All camera devices with at least 1 android.request.maxNumInputStreams will have at least one available input format. The camera device will support the following map of formats, if its dependent capability (android.request.availableCapabilities) is supported: Input Format | Output Format | Capability :-------------------------------------------------|:--------------------------------------------------|:---------- {@link android.graphics.ImageFormat#PRIVATE} | {@link android.graphics.ImageFormat#JPEG} | PRIVATE_REPROCESSING {@link android.graphics.ImageFormat#PRIVATE} | {@link android.graphics.ImageFormat#YUV_420_888} | PRIVATE_REPROCESSING {@link android.graphics.ImageFormat#YUV_420_888} | {@link android.graphics.ImageFormat#JPEG} | YUV_REPROCESSING {@link android.graphics.ImageFormat#YUV_420_888} | {@link android.graphics.ImageFormat#YUV_420_888} | YUV_REPROCESSING PRIVATE refers to a device-internal format that is not directly application-visible. A PRIVATE input surface can be acquired by {@link android.media.ImageReader#newInstance} with {@link android.graphics.ImageFormat#PRIVATE} as the format. For a PRIVATE_REPROCESSING-capable camera device, using the PRIVATE format as either input or output will never hurt maximum frame rate (i.e. {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration getOutputStallDuration(ImageFormat.PRIVATE, size)} is always 0), Attempting to configure an input stream with output streams not listed as available in this map is not valid.

For the formats, see `system/core/include/system/graphics.h` for a definition of the image format enumerations. The PRIVATE format refers to the HAL_PIXEL_FORMAT_IMPLEMENTATION_DEFINED format. The HAL could determine the actual format by using the gralloc usage flags. For ZSL use case in particular, the HAL could choose appropriate format (partially processed YUV or RAW based format) by checking the format and GRALLOC_USAGE_HW_CAMERA_ZSL. See camera3.h for more details. This value is encoded as a variable-size array-of-arrays. The inner array always contains `[format, length, ...]` where `...` has `length` elements. An inner array is followed by another inner array if the total metadata entry size hasn't yet been exceeded. A code sample to read/write this encoding (with a device that supports reprocessing IMPLEMENTATION_DEFINED to YUV_420_888, and JPEG, and reprocessing YUV_420_888 to YUV_420_888 and JPEG): // reading int32_t* contents = &entry.i32[0]; for (size_t i = 0; i < entry.count; ) { int32_t format = contents[i++]; int32_t length = contents[i++]; int32_t output_formats[length]; memcpy(&output_formats[0], &contents[i], length * sizeof(int32_t)); i += length; } // writing (static example, PRIVATE_REPROCESSING + YUV_REPROCESSING) int32_t[] contents = { IMPLEMENTATION_DEFINED, 2, YUV_420_888, BLOB, YUV_420_888, 2, YUV_420_888, BLOB, }; update_camera_metadata_entry(metadata, index, &contents[0], sizeof(contents)/sizeof(contents[0]), &updated_entry); If the HAL claims to support any of the capabilities listed in the above details, then it must also support all the input-output combinations listed for that capability. It can optionally support additional formats if it so chooses. n 4 OUTPUT INPUT The available stream configurations that this camera device supports (i.e. format, width, height, output/input stream).

The configurations are listed as `(format, width, height, input?)` tuples. For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide. Please reference the documentation for the image data destination to check if it limits the maximum size for image data. Not all output formats may be supported in a configuration with an input stream of a particular format. For more details, see android.scaler.availableInputOutputFormatsMap. The following table describes the minimum required output stream configurations based on the hardware level (android.info.supportedHardwareLevel): Format | Size | Hardware Level | Notes :-------------:|:--------------------------------------------:|:--------------:|:--------------: JPEG | android.sensor.info.activeArraySize | Any | JPEG | 1920x1080 (1080p) | Any | if 1080p <= activeArraySize JPEG | 1280x720 (720) | Any | if 720p <= activeArraySize JPEG | 640x480 (480p) | Any | if 480p <= activeArraySize JPEG | 320x240 (240p) | Any | if 240p <= activeArraySize YUV_420_888 | all output sizes available for JPEG | FULL | YUV_420_888 | all output sizes available for JPEG, up to the maximum video size | LIMITED | IMPLEMENTATION_DEFINED | same as YUV_420_888 | Any | Refer to android.request.availableCapabilities for additional mandatory stream configurations on a per-capability basis.

It is recommended (but not mandatory) to also include half/quarter of sensor maximum resolution for JPEG formats (regardless of hardware level). (The following is a rewording of the above required table): For JPEG format, the sizes may be restricted by below conditions: * The HAL may choose the aspect ratio of each Jpeg size to be one of well known ones (e.g. 4:3, 16:9, 3:2 etc.). If the sensor maximum resolution (defined by android.sensor.info.activeArraySize) has an aspect ratio other than these, it does not have to be included in the supported JPEG sizes. * Some hardware JPEG encoders may have pixel boundary alignment requirements, such as the dimensions being a multiple of 16. Therefore, the maximum JPEG size may be smaller than sensor maximum resolution. However, the largest JPEG size must be as close as possible to the sensor maximum resolution given above constraints. It is required that after aspect ratio adjustments, additional size reduction due to other issues must be less than 3% in area. For example, if the sensor maximum resolution is 3280x2464, if the maximum JPEG size has aspect ratio 4:3, the JPEG encoder alignment requirement is 16, the maximum JPEG size will be 3264x2448. For FULL capability devices (`android.info.supportedHardwareLevel == FULL`), the HAL must include all YUV_420_888 sizes that have JPEG sizes listed here as output streams. It must also include each below resolution if it is smaller than or equal to the sensor maximum resolution (for both YUV_420_888 and JPEG formats), as output streams: * 240p (320 x 240) * 480p (640 x 480) * 720p (1280 x 720) * 1080p (1920 x 1080) For LIMITED capability devices (`android.info.supportedHardwareLevel == LIMITED`), the HAL only has to list up to the maximum video size supported by the device. Regardless of hardware level, every output resolution available for YUV_420_888 must also be available for IMPLEMENTATION_DEFINED. This supercedes the following fields, which are now deprecated: * availableFormats * available[Processed,Raw,Jpeg]Sizes 4 n This lists the minimum frame duration for each format/size combination. (format, width, height, ns) x n

This should correspond to the frame duration when only that stream is active, with all processing (typically in android.*.mode) set to either OFF or FAST. When multiple streams are used in a request, the minimum frame duration will be max(individual stream min durations). The minimum frame duration of a stream (of a particular format, size) is the same regardless of whether the stream is input or output. See android.sensor.frameDuration and android.scaler.availableStallDurations for more details about calculating the max frame rate. (Keep in sync with {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration})

4 n This lists the maximum stall duration for each output format/size combination. (format, width, height, ns) x n

A stall duration is how much extra time would get added to the normal minimum frame duration for a repeating request that has streams with non-zero stall. For example, consider JPEG captures which have the following characteristics: * JPEG streams act like processed YUV streams in requests for which they are not included; in requests in which they are directly referenced, they act as JPEG streams. This is because supporting a JPEG stream requires the underlying YUV data to always be ready for use by a JPEG encoder, but the encoder will only be used (and impact frame duration) on requests that actually reference a JPEG stream. * The JPEG processor can run concurrently to the rest of the camera pipeline, but cannot process more than 1 capture at a time. In other words, using a repeating YUV request would result in a steady frame rate (let's say it's 30 FPS). If a single JPEG request is submitted periodically, the frame rate will stay at 30 FPS (as long as we wait for the previous JPEG to return each time). If we try to submit a repeating YUV + JPEG request, then the frame rate will drop from 30 FPS. In general, submitting a new request with a non-0 stall time stream will _not_ cause a frame rate drop unless there are still outstanding buffers for that stream from previous requests. Submitting a repeating request with streams (call this `S`) is the same as setting the minimum frame duration from the normal minimum frame duration corresponding to `S`, added with the maximum stall duration for `S`. If interleaving requests with and without a stall duration, a request will stall by the maximum of the remaining times for each can-stall stream with outstanding buffers. This means that a stalling request will not have an exposure start until the stall has completed. This should correspond to the stall duration when only that stream is active, with all processing (typically in android.*.mode) set to FAST or OFF. Setting any of the processing modes to HIGH_QUALITY effectively results in an indeterminate stall duration for all streams in a request (the regular stall calculation rules are ignored). The following formats may always have a stall duration: * {@link android.graphics.ImageFormat#JPEG} * {@link android.graphics.ImageFormat#RAW_SENSOR} The following formats will never have a stall duration: * {@link android.graphics.ImageFormat#YUV_420_888} * {@link android.graphics.ImageFormat#RAW10} All other formats may or may not have an allowed stall duration on a per-capability basis; refer to android.request.availableCapabilities for more details. See android.sensor.frameDuration for more information about calculating the max frame rate (absent stalls). (Keep up to date with {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} )

If possible, it is recommended that all non-JPEG formats (such as RAW16) should not have a stall duration. RAW10, RAW12, RAW_OPAQUE and IMPLEMENTATION_DEFINED must not have stall durations. The available stream configurations that this camera device supports; also includes the minimum frame durations and the stall durations for each format/size combination.

All camera devices will support sensor maximum resolution (defined by android.sensor.info.activeArraySize) for the JPEG format. For a given use case, the actual maximum supported resolution may be lower than what is listed here, depending on the destination Surface for the image data. For example, for recording video, the video encoder chosen may have a maximum size limit (e.g. 1080p) smaller than what the camera (e.g. maximum resolution is 3264x2448) can provide. Please reference the documentation for the image data destination to check if it limits the maximum size for image data. The following table describes the minimum required output stream configurations based on the hardware level (android.info.supportedHardwareLevel): Format | Size | Hardware Level | Notes :-------------------------------------------------:|:--------------------------------------------:|:--------------:|:--------------: {@link android.graphics.ImageFormat#JPEG} | android.sensor.info.activeArraySize | Any | {@link android.graphics.ImageFormat#JPEG} | 1920x1080 (1080p) | Any | if 1080p <= activeArraySize {@link android.graphics.ImageFormat#JPEG} | 1280x720 (720) | Any | if 720p <= activeArraySize {@link android.graphics.ImageFormat#JPEG} | 640x480 (480p) | Any | if 480p <= activeArraySize {@link android.graphics.ImageFormat#JPEG} | 320x240 (240p) | Any | if 240p <= activeArraySize {@link android.graphics.ImageFormat#YUV_420_888} | all output sizes available for JPEG | FULL | {@link android.graphics.ImageFormat#YUV_420_888} | all output sizes available for JPEG, up to the maximum video size | LIMITED | {@link android.graphics.ImageFormat#PRIVATE} | same as YUV_420_888 | Any | Refer to android.request.availableCapabilities and {@link android.hardware.camera2.CameraDevice#createCaptureSession} for additional mandatory stream configurations on a per-capability basis.

Do not set this property directly (it is synthetic and will not be available at the HAL layer); set the android.scaler.availableStreamConfigurations instead. Not all output formats may be supported in a configuration with an input stream of a particular format. For more details, see android.scaler.availableInputOutputFormatsMap. It is recommended (but not mandatory) to also include half/quarter of sensor maximum resolution for JPEG formats (regardless of hardware level). (The following is a rewording of the above required table): The HAL must include sensor maximum resolution (defined by android.sensor.info.activeArraySize). For FULL capability devices (`android.info.supportedHardwareLevel == FULL`), the HAL must include all YUV_420_888 sizes that have JPEG sizes listed here as output streams. It must also include each below resolution if it is smaller than or equal to the sensor maximum resolution (for both YUV_420_888 and JPEG formats), as output streams: * 240p (320 x 240) * 480p (640 x 480) * 720p (1280 x 720) * 1080p (1920 x 1080) For LIMITED capability devices (`android.info.supportedHardwareLevel == LIMITED`), the HAL only has to list up to the maximum video size supported by the device. Regardless of hardware level, every output resolution available for YUV_420_888 must also be available for IMPLEMENTATION_DEFINED. This supercedes the following fields, which are now deprecated: * availableFormats * available[Processed,Raw,Jpeg]Sizes CENTER_ONLY The camera device only supports centered crop regions. FREEFORM The camera device supports arbitrarily chosen crop regions. The crop type that this camera device supports.

When passing a non-centered crop region (android.scaler.cropRegion) to a camera device that only supports CENTER_ONLY cropping, the camera device will move the crop region to the center of the sensor active array (android.sensor.info.activeArraySize) and keep the crop region width and height unchanged. The camera device will return the final used crop region in metadata result android.scaler.cropRegion. Camera devices that support FREEFORM cropping will support any crop region that is inside of the active array. The camera device will apply the same crop region and return the final used crop region in capture result metadata android.scaler.cropRegion. FULL capability devices (android.info.supportedHardwareLevel `==` FULL) will support FREEFORM cropping. LEGACY capability devices will only support CENTER_ONLY cropping.

Duration each pixel is exposed to light. Nanoseconds android.sensor.info.exposureTimeRange

If the sensor can't expose this exact duration, it will shorten the duration exposed to the nearest possible value (rather than expose longer). The final exposure time used will be available in the output capture result. This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

Duration from start of frame exposure to start of next frame exposure. Nanoseconds See android.sensor.info.maxFrameDuration, android.scaler.streamConfigurationMap. The duration is capped to `max(duration, exposureTime + overhead)`.

The maximum frame rate that can be supported by a camera subsystem is a function of many factors: * Requested resolutions of output image streams * Availability of binning / skipping modes on the imager * The bandwidth of the imager interface * The bandwidth of the various ISP processing blocks Since these factors can vary greatly between different ISPs and sensors, the camera abstraction tries to represent the bandwidth restrictions with as simple a model as possible. The model presented has the following characteristics: * The image sensor is always configured to output the smallest resolution possible given the application's requested output stream sizes. The smallest resolution is defined as being at least as large as the largest requested output stream size; the camera pipeline must never digitally upsample sensor data when the crop region covers the whole sensor. In general, this means that if only small output stream resolutions are configured, the sensor can provide a higher frame rate. * Since any request may use any or all the currently configured output streams, the sensor and ISP must be configured to support scaling a single capture to all the streams at the same time. This means the camera pipeline must be ready to produce the largest requested output size without any delay. Therefore, the overall frame rate of a given configured stream set is governed only by the largest requested stream resolution. * Using more than one output stream in a request does not affect the frame duration. * Certain format-streams may need to do additional background processing before data is consumed/produced by that stream. These processors can run concurrently to the rest of the camera pipeline, but cannot process more than 1 capture at a time. The necessary information for the application, given the model above, is provided via the android.scaler.streamConfigurationMap field using {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration}. These are used to determine the maximum frame rate / minimum frame duration that is possible for a given stream configuration. Specifically, the application can use the following rules to determine the minimum frame duration it can request from the camera device: 1. Let the set of currently configured input/output streams be called `S`. 1. Find the minimum frame durations for each stream in `S`, by looking it up in android.scaler.streamConfigurationMap using {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration} (with its respective size/format). Let this set of frame durations be called `F`. 1. For any given request `R`, the minimum frame duration allowed for `R` is the maximum out of all values in `F`. Let the streams used in `R` be called `S_r`. If none of the streams in `S_r` have a stall time (listed in {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration} using its respective size/format), then the frame duration in `F` determines the steady state frame rate that the application will get if it uses `R` as a repeating request. Let this special kind of request be called `Rsimple`. A repeating request `Rsimple` can be _occasionally_ interleaved by a single capture of a new request `Rstall` (which has at least one in-use stream with a non-0 stall time) and if `Rstall` has the same minimum frame duration this will not cause a frame rate loss if all buffers from the previous `Rstall` have already been delivered. For more details about stalling, see {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputStallDuration}. This control is only effective if android.control.aeMode or android.control.mode is set to OFF; otherwise the auto-exposure algorithm will override this value.

For more details about stalling, see android.scaler.availableStallDurations. The amount of gain applied to sensor data before processing. ISO arithmetic units android.sensor.info.sensitivityRange

The sensitivity is the standard ISO sensitivity value, as defined in ISO 12232:2006. The sensitivity must be within android.sensor.info.sensitivityRange, and if if it less than android.sensor.maxAnalogSensitivity, the camera device is guaranteed to use only analog amplification for applying the gain. If the camera device cannot apply the exact sensitivity requested, it will reduce the gain to the nearest supported value. The final sensitivity used will be available in the output capture result.

ISO 12232:2006 REI method is acceptable. 4 The area of the image sensor which corresponds to active pixels after any geometric distortion correction has been applied. Pixel coordinates on the image sensor

This is the rectangle representing the size of the active region of the sensor (i.e. the region that actually receives light from the scene) after any geometric correction has been applied, and should be treated as the maximum size in pixels of any of the image output formats aside from the raw formats. This rectangle is defined relative to the full pixel array; (0,0) is the top-left of the full pixel array, and the size of the full pixel array is given by android.sensor.info.pixelArraySize. The coordinate system for most other keys that list pixel coordinates, including android.scaler.cropRegion, is defined relative to the active array rectangle given in this field, with `(0, 0)` being the top-left of this rectangle. The active array may be smaller than the full pixel array, since the full array may include black calibration pixels or other inactive regions, and geometric correction resulting in scaling or cropping may have been applied.

This array contains `(xmin, ymin, width, height)`. The `(xmin, ymin)` must be >= `(0,0)`. The `(width, height)` must be <= `android.sensor.info.pixelArraySize`. 4 The area of the image sensor which corresponds to active pixels prior to the application of any geometric distortion correction. Pixel coordinates on the image sensor

This is the rectangle representing the size of the active region of the sensor (i.e. the region that actually receives light from the scene) before any geometric correction has been applied, and should be treated as the active region rectangle for any of the raw formats. All metadata associated with raw processing (e.g. the lens shading correction map, and radial distortion fields) treats the top, left of this rectangle as the origin, (0,0). The size of this region determines the maximum field of view and the maximum number of pixels that an image from this sensor can contain, prior to the application of geometric distortion correction. The effective maximum pixel dimensions of a post-distortion-corrected image is given by the android.sensor.info.activeArraySize field, and the effective maximum field of view for a post-distortion-corrected image can be calculated by applying the geometric distortion correction fields to this rectangle, and cropping to the rectangle given in android.sensor.info.activeArraySize. E.g. to calculate position of a pixel, (x,y), in a processed YUV output image with the dimensions in android.sensor.info.activeArraySize given the position of a pixel, (x', y'), in the raw pixel array with dimensions give in android.sensor.info.pixelArraySize: 1. Choose a pixel (x', y') within the active array region of the raw buffer given in android.sensor.info.preCorrectionActiveArraySize, otherwise this pixel is considered to be outside of the FOV, and will not be shown in the processed output image. 1. Apply geometric distortion correction to get the post-distortion pixel coordinate, (x_i, y_i). When applying geometric correction metadata, note that metadata for raw buffers is defined relative to the top, left of the android.sensor.info.preCorrectionActiveArraySize rectangle. 1. If the resulting corrected pixel coordinate is within the region given in android.sensor.info.activeArraySize, then the position of this pixel in the processed output image buffer is `(x_i - activeArray.left, y_i - activeArray.top)`, when the top, left coordinate of that buffer is treated as (0, 0). Thus, for pixel x',y' = (25, 25) on a sensor where android.sensor.info.pixelArraySize is (100,100), android.sensor.info.preCorrectionActiveArraySize is (10, 10, 100, 100), android.sensor.info.activeArraySize is (20, 20, 80, 80), and the geometric distortion correction doesn't change the pixel coordinate, the resulting pixel selected in pixel coordinates would be x,y = (25, 25) relative to the top,left of the raw buffer with dimensions given in android.sensor.info.pixelArraySize, and would be (5, 5) relative to the top,left of post-processed YUV output buffer with dimensions given in android.sensor.info.activeArraySize. The currently supported fields that correct for geometric distortion are: 1. android.lens.radialDistortion. If all of the geometric distortion fields are no-ops, this rectangle will be the same as the post-distortion-corrected rectangle given in android.sensor.info.activeArraySize. This rectangle is defined relative to the full pixel array; (0,0) is the top-left of the full pixel array, and the size of the full pixel array is given by android.sensor.info.pixelArraySize. The pre-correction active array may be smaller than the full pixel array, since the full array may include black calibration pixels or other inactive regions.

This array contains `(xmin, ymin, width, height)`. The `(xmin, ymin)` must be >= `(0,0)`. The `(width, height)` must be <= `android.sensor.info.pixelArraySize`. 2 Range of sensitivities for android.sensor.sensitivity supported by this camera device. Min <= 100, Max >= 800

The values are the standard ISO sensitivity values, as defined in ISO 12232:2006.

RGGB GRBG GBRG BGGR RGB Sensor is not Bayer; output has 3 16-bit values for each pixel, instead of just 1 16-bit value per pixel. The arrangement of color filters on sensor; represents the colors in the top-left 2x2 section of the sensor, in reading order. 2 The range of image exposure times for android.sensor.exposureTime supported by this camera device. Nanoseconds The minimum exposure time will be less than 100 us. For FULL capability devices (android.info.supportedHardwareLevel == FULL), the maximum exposure time will be greater than 100ms. For FULL capability devices (android.info.supportedHardwareLevel == FULL), The maximum of the range SHOULD be at least 1 second (1e9), MUST be at least 100ms. The maximum possible frame duration (minimum frame rate) for android.sensor.frameDuration that is supported this camera device. Nanoseconds For FULL capability devices (android.info.supportedHardwareLevel == FULL), at least 100ms.

Attempting to use frame durations beyond the maximum will result in the frame duration being clipped to the maximum. See that control for a full definition of frame durations. Refer to {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration} for the minimum frame duration values.

For FULL capability devices (android.info.supportedHardwareLevel == FULL), The maximum of the range SHOULD be at least 1 second (1e9), MUST be at least 100ms (100e6). android.sensor.info.maxFrameDuration must be greater or equal to the android.sensor.info.exposureTimeRange max value (since exposure time overrides frame duration). Available minimum frame durations for JPEG must be no greater than that of the YUV_420_888/IMPLEMENTATION_DEFINED minimum frame durations (for that respective size). Since JPEG processing is considered offline and can take longer than a single uncompressed capture, refer to android.scaler.availableStallDurations for details about encoding this scenario. 2 The physical dimensions of the full pixel array. Millimeters

This is the physical size of the sensor pixel array defined by android.sensor.info.pixelArraySize.

Needed for FOV calculation for old API 2 Dimensions of the full pixel array, possibly including black calibration pixels. Pixels

The pixel count of the full pixel array of the image sensor, which covers android.sensor.info.physicalSize area. This represents the full pixel dimensions of the raw buffers produced by this sensor. If a camera device supports raw sensor formats, either this or android.sensor.info.preCorrectionActiveArraySize is the maximum dimensions for the raw output formats listed in android.scaler.streamConfigurationMap (this depends on whether or not the image sensor returns buffers containing pixels that are not part of the active array region for blacklevel calibration or other purposes). Some parts of the full pixel array may not receive light from the scene, or be otherwise inactive. The android.sensor.info.preCorrectionActiveArraySize key defines the rectangle of active pixels that will be included in processed image formats.

Maximum raw value output by sensor. > 255 (8-bit output)

This specifies the fully-saturated encoding level for the raw sample values from the sensor. This is typically caused by the sensor becoming highly non-linear or clipping. The minimum for each channel is specified by the offset in the android.sensor.blackLevelPattern key. The white level is typically determined either by sensor bit depth (8-14 bits is expected), or by the point where the sensor response becomes too non-linear to be useful. The default value for this is maximum representable value for a 16-bit raw sample (2^16 - 1).

The full bit depth of the sensor must be available in the raw data, so the value for linear sensors should not be significantly lower than maximum raw value supported, i.e. 2^(sensor bits per pixel). UNKNOWN Timestamps from android.sensor.timestamp are in nanoseconds and monotonic, but can not be compared to timestamps from other subsystems (e.g. accelerometer, gyro etc.), or other instances of the same or different camera devices in the same system. Timestamps between streams and results for a single camera instance are comparable, and the timestamps for all buffers and the result metadata generated by a single capture are identical. REALTIME Timestamps from android.sensor.timestamp are in the same timebase as {@link android.os.SystemClock#elapsedRealtimeNanos}, and they can be compared to other timestamps using that base. The time base source for sensor capture start timestamps.

The timestamps provided for captures are always in nanoseconds and monotonic, but may not based on a time source that can be compared to other system time sources. This characteristic defines the source for the timestamps, and therefore whether they can be compared against other system time sources/timestamps.

FALSE TRUE Whether the RAW images output from this camera device are subject to lens shading correction.

If TRUE, all images produced by the camera device in the RAW image formats will have lens shading correction already applied to it. If FALSE, the images will not be adjusted for lens shading correction. See android.request.maxNumOutputRaw for a list of RAW image formats. This key will be `null` for all devices do not report this information. Devices with RAW capability will always report this information in this key.

DAYLIGHT FLUORESCENT TUNGSTEN Incandescent light FLASH FINE_WEATHER CLOUDY_WEATHER SHADE DAYLIGHT_FLUORESCENT D 5700 - 7100K DAY_WHITE_FLUORESCENT N 4600 - 5400K COOL_WHITE_FLUORESCENT W 3900 - 4500K WHITE_FLUORESCENT WW 3200 - 3700K STANDARD_A STANDARD_B STANDARD_C D55 D65 D75 D50 ISO_STUDIO_TUNGSTEN The standard reference illuminant used as the scene light source when calculating the android.sensor.colorTransform1, android.sensor.calibrationTransform1, and android.sensor.forwardMatrix1 matrices.

The values in this key correspond to the values defined for the EXIF LightSource tag. These illuminants are standard light sources that are often used calibrating camera devices. If this key is present, then android.sensor.colorTransform1, android.sensor.calibrationTransform1, and android.sensor.forwardMatrix1 will also be present. Some devices may choose to provide a second set of calibration information for improved quality, including android.sensor.referenceIlluminant2 and its corresponding matrices.

The first reference illuminant (android.sensor.referenceIlluminant1) and corresponding matrices must be present to support the RAW capability and DNG output. When producing raw images with a color profile that has only been calibrated against a single light source, it is valid to omit android.sensor.referenceIlluminant2 along with the android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 matrices. If only android.sensor.referenceIlluminant1 is included, it should be chosen so that it is representative of typical scene lighting. In general, D50 or DAYLIGHT will be chosen for this case. If both android.sensor.referenceIlluminant1 and android.sensor.referenceIlluminant2 are included, they should be chosen to represent the typical range of scene lighting conditions. In general, low color temperature illuminant such as Standard-A will be chosen for the first reference illuminant and a higher color temperature illuminant such as D65 will be chosen for the second reference illuminant. The standard reference illuminant used as the scene light source when calculating the android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 matrices. Any value listed in android.sensor.referenceIlluminant1

See android.sensor.referenceIlluminant1 for more details. If this key is present, then android.sensor.colorTransform2, android.sensor.calibrationTransform2, and android.sensor.forwardMatrix2 will also be present.

3 3 A per-device calibration transform matrix that maps from the reference sensor colorspace to the actual device sensor colorspace.

This matrix is used to correct for per-device variations in the sensor colorspace, and is used for processing raw buffer data. The matrix is expressed as a 3x3 matrix in row-major-order, and contains a per-device calibration transform that maps colors from reference sensor color space (i.e. the "golden module" colorspace) into this camera device's native sensor color space under the first reference illuminant (android.sensor.referenceIlluminant1).

3 3 A per-device calibration transform matrix that maps from the reference sensor colorspace to the actual device sensor colorspace (this is the colorspace of the raw buffer data).

This matrix is used to correct for per-device variations in the sensor colorspace, and is used for processing raw buffer data. The matrix is expressed as a 3x3 matrix in row-major-order, and contains a per-device calibration transform that maps colors from reference sensor color space (i.e. the "golden module" colorspace) into this camera device's native sensor color space under the second reference illuminant (android.sensor.referenceIlluminant2). This matrix will only be present if the second reference illuminant is present.

3 3 A matrix that transforms color values from CIE XYZ color space to reference sensor color space.

This matrix is used to convert from the standard CIE XYZ color space to the reference sensor colorspace, and is used when processing raw buffer data. The matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps colors from the CIE XYZ color space to the reference sensor color space (i.e. the "golden module" colorspace) under the first reference illuminant (android.sensor.referenceIlluminant1). The white points chosen in both the reference sensor color space and the CIE XYZ colorspace when calculating this transform will match the standard white point for the first reference illuminant (i.e. no chromatic adaptation will be applied by this transform).

3 3 A matrix that transforms color values from CIE XYZ color space to reference sensor color space.

This matrix is used to convert from the standard CIE XYZ color space to the reference sensor colorspace, and is used when processing raw buffer data. The matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps colors from the CIE XYZ color space to the reference sensor color space (i.e. the "golden module" colorspace) under the second reference illuminant (android.sensor.referenceIlluminant2). The white points chosen in both the reference sensor color space and the CIE XYZ colorspace when calculating this transform will match the standard white point for the second reference illuminant (i.e. no chromatic adaptation will be applied by this transform). This matrix will only be present if the second reference illuminant is present.

3 3 A matrix that transforms white balanced camera colors from the reference sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.

This matrix is used to convert to the standard CIE XYZ colorspace, and is used when processing raw buffer data. This matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps white balanced colors from the reference sensor color space to the CIE XYZ color space with a D50 white point. Under the first reference illuminant (android.sensor.referenceIlluminant1) this matrix is chosen so that the standard white point for this reference illuminant in the reference sensor colorspace is mapped to D50 in the CIE XYZ colorspace.

3 3 A matrix that transforms white balanced camera colors from the reference sensor colorspace to the CIE XYZ colorspace with a D50 whitepoint.

This matrix is used to convert to the standard CIE XYZ colorspace, and is used when processing raw buffer data. This matrix is expressed as a 3x3 matrix in row-major-order, and contains a color transform matrix that maps white balanced colors from the reference sensor color space to the CIE XYZ color space with a D50 white point. Under the second reference illuminant (android.sensor.referenceIlluminant2) this matrix is chosen so that the standard white point for this reference illuminant in the reference sensor colorspace is mapped to D50 in the CIE XYZ colorspace. This matrix will only be present if the second reference illuminant is present.

Gain factor from electrons to raw units when ISO=100 4 A fixed black level offset for each of the color filter arrangement (CFA) mosaic channels. >= 0 for each.

This key specifies the zero light value for each of the CFA mosaic channels in the camera sensor. The maximal value output by the sensor is represented by the value in android.sensor.info.whiteLevel. The values are given in the same order as channels listed for the CFA layout key (see android.sensor.info.colorFilterArrangement), i.e. the nth value given corresponds to the black level offset for the nth color channel listed in the CFA.

The values are given in row-column scan order, with the first value corresponding to the element of the CFA in row=0, column=0. Maximum sensitivity that is implemented purely through analog gain.

For android.sensor.sensitivity values less than or equal to this, all applied gain must be analog. For values above this, the gain applied can be a mix of analog and digital.

Clockwise angle through which the output image needs to be rotated to be upright on the device screen in its native orientation. Degrees of clockwise rotation; always a multiple of 90 0, 90, 180, 270

Also defines the direction of rolling shutter readout, which is from top to bottom in the sensor's coordinate system.

3 The number of input samples for each dimension of android.sensor.profileHueSatMap. Hue >= 1, Saturation >= 2, Value >= 1

The number of input samples for the hue, saturation, and value dimension of android.sensor.profileHueSatMap. The order of the dimensions given is hue, saturation, value; where hue is the 0th element.

Time at start of exposure of first row of the image sensor active array, in nanoseconds. Nanoseconds > 0

The timestamps are also included in all image buffers produced for the same capture, and will be identical on all the outputs. When android.sensor.info.timestampSource `==` UNKNOWN, the timestamps measure time since an unspecified starting point, and are monotonically increasing. They can be compared with the timestamps for other captures from the same camera device, but are not guaranteed to be comparable to any other time source. When android.sensor.info.timestampSource `==` REALTIME, the timestamps measure time in the same timebase as {@link android.os.SystemClock#elapsedRealtimeNanos}, and they can be compared to other timestamps from other subsystems that are using that base. For reprocessing, the timestamp will match the start of exposure of the input image, i.e. {@link CaptureResult#SENSOR_TIMESTAMP the timestamp} in the TotalCaptureResult that was used to create the reprocess capture request.

All timestamps must be in reference to the kernel's CLOCK_BOOTTIME monotonic clock, which properly accounts for time spent asleep. This allows for synchronization with sensors that continue to operate while the system is otherwise asleep. If android.sensor.info.timestampSource `==` REALTIME, The timestamp must be synchronized with the timestamps from other sensor subsystems that are using the same timebase. For reprocessing, the input image's start of exposure can be looked up with android.sensor.timestamp from the metadata included in the capture request. The temperature of the sensor, sampled at the time exposure began for this frame. The thermal diode being queried should be inside the sensor PCB, or somewhere close to it. Celsius Optional. This value is missing if no temperature is available. 3 The estimated camera neutral color in the native sensor colorspace at the time of capture.

This value gives the neutral color point encoded as an RGB value in the native sensor color space. The neutral color point indicates the currently estimated white point of the scene illumination. It can be used to interpolate between the provided color transforms when processing raw sensor data. The order of the values is R, G, B; where R is in the lowest index.

2 CFA Channels Noise model coefficients for each CFA mosaic channel.

This key contains two noise model coefficients for each CFA channel corresponding to the sensor amplification (S) and sensor readout noise (O). These are given as pairs of coefficients for each channel in the same order as channels listed for the CFA layout key (see android.sensor.info.colorFilterArrangement). This is represented as an array of Pair<Double, Double>, where the first member of the Pair at index n is the S coefficient and the second member is the O coefficient for the nth color channel in the CFA. These coefficients are used in a two parameter noise model to describe the amount of noise present in the image for each CFA channel. The noise model used here is: N(x) = sqrt(Sx + O) Where x represents the recorded signal of a CFA channel normalized to the range [0, 1], and S and O are the noise model coeffiecients for that channel. A more detailed description of the noise model can be found in the Adobe DNG specification for the NoiseProfile tag.

For a CFA layout of RGGB, the list of coefficients would be given as an array of doubles S0,O0,S1,O1,..., where S0 and O0 are the coefficients for the red channel, S1 and O1 are the coefficients for the first green channel, etc. hue_samples saturation_samples value_samples 3 A mapping containing a hue shift, saturation scale, and value scale for each pixel. The hue shift is given in degrees; saturation and value scale factors are unitless and are between 0 and 1 inclusive

hue_samples, saturation_samples, and value_samples are given in android.sensor.profileHueSatMapDimensions. Each entry of this map contains three floats corresponding to the hue shift, saturation scale, and value scale, respectively; where the hue shift has the lowest index. The map entries are stored in the key in nested loop order, with the value divisions in the outer loop, the hue divisions in the middle loop, and the saturation divisions in the inner loop. All zero input saturation entries are required to have a value scale factor of 1.0.

samples 2 A list of x,y samples defining a tone-mapping curve for gamma adjustment. Each sample has an input range of `[0, 1]` and an output range of `[0, 1]`. The first sample is required to be `(0, 0)`, and the last sample is required to be `(1, 1)`.

This key contains a default tone curve that can be applied while processing the image as a starting point for user adjustments. The curve is specified as a list of value pairs in linear gamma. The curve is interpolated using a cubic spline.

The worst-case divergence between Bayer green channels. >= 0

This value is an estimate of the worst case split between the Bayer green channels in the red and blue rows in the sensor color filter array. The green split is calculated as follows: 1. A 5x5 pixel (or larger) window W within the active sensor array is chosen. The term 'pixel' here is taken to mean a group of 4 Bayer mosaic channels (R, Gr, Gb, B). The location and size of the window chosen is implementation defined, and should be chosen to provide a green split estimate that is both representative of the entire image for this camera sensor, and can be calculated quickly. 1. The arithmetic mean of the green channels from the red rows (mean_Gr) within W is computed. 1. The arithmetic mean of the green channels from the blue rows (mean_Gb) within W is computed. 1. The maximum ratio R of the two means is computed as follows: `R = max((mean_Gr + 1)/(mean_Gb + 1), (mean_Gb + 1)/(mean_Gr + 1))` The ratio R is the green split divergence reported for this property, which represents how much the green channels differ in the mosaic pattern. This value is typically used to determine the treatment of the green mosaic channels when demosaicing. The green split value can be roughly interpreted as follows: * R < 1.03 is a negligible split (<3% divergence). * 1.20 <= R >= 1.03 will require some software correction to avoid demosaic errors (3-20% divergence). * R > 1.20 will require strong software correction to produce a usuable image (>20% divergence).

The green split given may be a static value based on prior characterization of the camera sensor using the green split calculation method given here over a large, representative, sample set of images. Other methods of calculation that produce equivalent results, and can be interpreted in the same manner, may be used. 4 A pixel `[R, G_even, G_odd, B]` that supplies the test pattern when android.sensor.testPatternMode is SOLID_COLOR.

Each color channel is treated as an unsigned 32-bit integer. The camera device then uses the most significant X bits that correspond to how many bits are in its Bayer raw sensor output. For example, a sensor with RAW10 Bayer output would use the 10 most significant bits from each color channel.

OFF No test pattern mode is used, and the camera device returns captures from the image sensor. This is the default if the key is not set. SOLID_COLOR Each pixel in `[R, G_even, G_odd, B]` is replaced by its respective color channel provided in android.sensor.testPatternData. For example: android.testPatternData = [0, 0xFFFFFFFF, 0xFFFFFFFF, 0] All green pixels are 100% green. All red/blue pixels are black. android.testPatternData = [0xFFFFFFFF, 0, 0xFFFFFFFF, 0] All red pixels are 100% red. Only the odd green pixels are 100% green. All blue pixels are 100% black. COLOR_BARS All pixel data is replaced with an 8-bar color pattern. The vertical bars (left-to-right) are as follows: * 100% white * yellow * cyan * green * magenta * red * blue * black In general the image would look like the following: W Y C G M R B K W Y C G M R B K W Y C G M R B K W Y C G M R B K W Y C G M R B K . . . . . . . . . . . . . . . . . . . . . . . . (B = Blue, K = Black) Each bar should take up 1/8 of the sensor pixel array width. When this is not possible, the bar size should be rounded down to the nearest integer and the pattern can repeat on the right side. Each bar's height must always take up the full sensor pixel array height. Each pixel in this test pattern must be set to either 0% intensity or 100% intensity. COLOR_BARS_FADE_TO_GRAY The test pattern is similar to COLOR_BARS, except that each bar should start at its specified color at the top, and fade to gray at the bottom. Furthermore each bar is further subdivided into a left and right half. The left half should have a smooth gradient, and the right half should have a quantized gradient. In particular, the right half's should consist of blocks of the same color for 1/16th active sensor pixel array width. The least significant bits in the quantized gradient should be copied from the most significant bits of the smooth gradient. The height of each bar should always be a multiple of 128. When this is not the case, the pattern should repeat at the bottom of the image. PN9 All pixel data is replaced by a pseudo-random sequence generated from a PN9 512-bit sequence (typically implemented in hardware with a linear feedback shift register). The generator should be reset at the beginning of each frame, and thus each subsequent raw frame with this test pattern should be exactly the same as the last. CUSTOM1 The first custom test pattern. All custom patterns that are available only on this camera device are at least this numeric value. All of the custom test patterns will be static (that is the raw image must not vary from frame to frame). When enabled, the sensor sends a test pattern instead of doing a real exposure from the camera. android.sensor.availableTestPatternModes

When a test pattern is enabled, all manual sensor controls specified by android.sensor.* will be ignored. All other controls should work as normal. For example, if manual flash is enabled, flash firing should still occur (and that the test pattern remain unmodified, since the flash would not actually affect it). Defaults to OFF.

All test patterns are specified in the Bayer domain. The HAL may choose to substitute test patterns from the sensor with test patterns from on-device memory. In that case, it should be indistinguishable to the ISP whether the data came from the sensor interconnect bus (such as CSI2) or memory. n List of sensor test pattern modes for android.sensor.testPatternMode supported by this camera device. Any value listed in android.sensor.testPatternMode

Defaults to OFF, and always includes OFF if defined.

All custom modes must be >= CUSTOM1. Duration between the start of first row exposure and the start of last row exposure. Nanoseconds >= 0 and < {@link android.hardware.camera2.params.StreamConfigurationMap#getOutputMinFrameDuration}.

This is the exposure time skew between the first and last row exposure start times. The first row and the last row are the first and last rows inside of the android.sensor.info.activeArraySize. For typical camera sensors that use rolling shutters, this is also equivalent to the frame readout time.

The HAL must report `0` if the sensor is using global shutter, where all pixels begin exposure at the same time.

OFF No lens shading correction is applied. FAST Apply lens shading corrections, without slowing frame rate relative to sensor raw output HIGH_QUALITY Apply high-quality lens shading correction, at the cost of possibly reduced frame rate. Quality of lens shading correction applied to the image data. android.shading.availableModes

When set to OFF mode, no lens shading correction will be applied by the camera device, and an identity lens shading map data will be provided if `android.statistics.lensShadingMapMode == ON`. For example, for lens shading map with size of `[ 4, 3 ]`, the output android.statistics.lensShadingCorrectionMap for this case will be an identity map shown below: [ 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0, 1.0 ] When set to other modes, lens shading correction will be applied by the camera device. Applications can request lens shading map data by setting android.statistics.lensShadingMapMode to ON, and then the camera device will provide lens shading map data in android.statistics.lensShadingCorrectionMap; the returned shading map data will be the one applied by the camera device for this capture request. The shading map data may depend on the auto-exposure (AE) and AWB statistics, therefore the reliability of the map data may be affected by the AE and AWB algorithms. When AE and AWB are in AUTO modes(android.control.aeMode `!=` OFF and android.control.awbMode `!=` OFF), to get best results, it is recommended that the applications wait for the AE and AWB to be converged before using the returned shading map data.

Control the amount of shading correction applied to the images unitless: 1-10; 10 is full shading compensation n List of lens shading modes for android.shading.mode that are supported by this camera device. Any value listed in android.shading.mode

This list contains lens shading modes that can be set for the camera device. Camera devices that support the MANUAL_POST_PROCESSING capability will always list OFF and FAST mode. This includes all FULL level devices. LEGACY devices will always only support FAST mode.

HAL must support both FAST and HIGH_QUALITY if lens shading correction control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output.

OFF Do not include face detection statistics in capture results. SIMPLE Return face rectangle and confidence values only. FULL Return all face metadata. In this mode, face rectangles, scores, landmarks, and face IDs are all valid. Operating mode for the face detector unit. android.statistics.info.availableFaceDetectModes

Whether face detection is enabled, and whether it should output just the basic fields or the full set of fields.

SIMPLE mode must fill in android.statistics.faceRectangles and android.statistics.faceScores. FULL mode must also fill in android.statistics.faceIds, and android.statistics.faceLandmarks. OFF ON Operating mode for histogram generation OFF ON Operating mode for sharpness map generation OFF Hot pixel map production is disabled. ON Hot pixel map production is enabled. Operating mode for hot pixel map generation. android.statistics.info.availableHotPixelMapModes

If set to `true`, a hot pixel map is returned in android.statistics.hotPixelMap. If set to `false`, no hot pixel map will be returned.

n List of face detection modes for android.statistics.faceDetectMode that are supported by this camera device. Any value listed in android.statistics.faceDetectMode

OFF is always supported.

Number of histogram buckets supported >= 64 The maximum number of simultaneously detectable faces. 0 for cameras without available face detection; otherwise: `>=4` for LIMITED or FULL hwlevel devices or `>0` for LEGACY devices. Maximum value possible for a histogram bucket Maximum value possible for a sharpness map region. 2 Dimensions of the sharpness map Must be at least 32 x 32 n List of hot pixel map output modes for android.statistics.hotPixelMapMode that are supported by this camera device. Any value listed in android.statistics.hotPixelMapMode

If no hotpixel map output is available for this camera device, this will contain only `false`. ON is always supported on devices with the RAW capability.

n List of lens shading map output modes for android.statistics.lensShadingMapMode that are supported by this camera device. Any value listed in android.statistics.lensShadingMapMode

If no lens shading map output is available for this camera device, this key will contain only OFF. ON is always supported on devices with the RAW capability. LEGACY mode devices will always only support OFF.

n List of unique IDs for detected faces.

Each detected face is given a unique ID that is valid for as long as the face is visible to the camera device. A face that leaves the field of view and later returns may be assigned a new ID. Only available if android.statistics.faceDetectMode == FULL

n 6 List of landmarks for detected faces.

The coordinate system is that of android.sensor.info.activeArraySize, with `(0, 0)` being the top-left pixel of the active array. Only available if android.statistics.faceDetectMode == FULL

n 4 List of the bounding rectangles for detected faces.

The coordinate system is that of android.sensor.info.activeArraySize, with `(0, 0)` being the top-left pixel of the active array. Only available if android.statistics.faceDetectMode != OFF

n List of the face confidence scores for detected faces 1-100

Only available if android.statistics.faceDetectMode != OFF.

The value should be meaningful (for example, setting 100 at all times is illegal). n List of the faces detected through camera face detection in this capture.

Only available if android.statistics.faceDetectMode `!=` OFF.

n 3 A 3-channel histogram based on the raw sensor data

The k'th bucket (0-based) covers the input range (with w = android.sensor.info.whiteLevel) of [ k * w/N, (k + 1) * w / N ). If only a monochrome sharpness map is supported, all channels should have the same data

n m 3 A 3-channel sharpness map, based on the raw sensor data

If only a monochrome sharpness map is supported, all channels should have the same data

The shading map is a low-resolution floating-point map that lists the coefficients used to correct for vignetting, for each Bayer color channel. Each gain factor is >= 1

The least shaded section of the image should have a gain factor of 1; all other sections should have gains above 1. When android.colorCorrection.mode = TRANSFORM_MATRIX, the map must take into account the colorCorrection settings. The shading map is for the entire active pixel array, and is not affected by the crop region specified in the request. Each shading map entry is the value of the shading compensation map over a specific pixel on the sensor. Specifically, with a (N x M) resolution shading map, and an active pixel array size (W x H), shading map entry (x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels. The map is assumed to be bilinearly interpolated between the sample points. The channel order is [R, Geven, Godd, B], where Geven is the green channel for the even rows of a Bayer pattern, and Godd is the odd rows. The shading map is stored in a fully interleaved format. The shading map should have on the order of 30-40 rows and columns, and must be smaller than 64x64. As an example, given a very small map defined as: width,height = [ 4, 3 ] values = [ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2, 1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3, 1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0, 1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2, 1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2, 1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ] The low-resolution scaling map images for each channel are (displayed using nearest-neighbor interpolation): ![Red lens shading map](android.statistics.lensShadingMap/red_shading.png) ![Green (even rows) lens shading map](android.statistics.lensShadingMap/green_e_shading.png) ![Green (odd rows) lens shading map](android.statistics.lensShadingMap/green_o_shading.png) ![Blue lens shading map](android.statistics.lensShadingMap/blue_shading.png) As a visualization only, inverting the full-color map to recover an image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives: ![Image of a uniform white wall (inverse shading map)](android.statistics.lensShadingMap/inv_shading.png)

4 n m The shading map is a low-resolution floating-point map that lists the coefficients used to correct for vignetting, for each Bayer color channel of RAW image data. Each gain factor is >= 1

The least shaded section of the image should have a gain factor of 1; all other sections should have gains above 1. When android.colorCorrection.mode = TRANSFORM_MATRIX, the map must take into account the colorCorrection settings. The shading map is for the entire active pixel array, and is not affected by the crop region specified in the request. Each shading map entry is the value of the shading compensation map over a specific pixel on the sensor. Specifically, with a (N x M) resolution shading map, and an active pixel array size (W x H), shading map entry (x,y) ϵ (0 ... N-1, 0 ... M-1) is the value of the shading map at pixel ( ((W-1)/(N-1)) * x, ((H-1)/(M-1)) * y) for the four color channels. The map is assumed to be bilinearly interpolated between the sample points. The channel order is [R, Geven, Godd, B], where Geven is the green channel for the even rows of a Bayer pattern, and Godd is the odd rows. The shading map is stored in a fully interleaved format, and its size is provided in the camera static metadata by android.lens.info.shadingMapSize. The shading map should have on the order of 30-40 rows and columns, and must be smaller than 64x64. As an example, given a very small map defined as: android.lens.info.shadingMapSize = [ 4, 3 ] android.statistics.lensShadingMap = [ 1.3, 1.2, 1.15, 1.2, 1.2, 1.2, 1.15, 1.2, 1.1, 1.2, 1.2, 1.2, 1.3, 1.2, 1.3, 1.3, 1.2, 1.2, 1.25, 1.1, 1.1, 1.1, 1.1, 1.0, 1.0, 1.0, 1.0, 1.0, 1.2, 1.3, 1.25, 1.2, 1.3, 1.2, 1.2, 1.3, 1.2, 1.15, 1.1, 1.2, 1.2, 1.1, 1.0, 1.2, 1.3, 1.15, 1.2, 1.3 ] The low-resolution scaling map images for each channel are (displayed using nearest-neighbor interpolation): ![Red lens shading map](android.statistics.lensShadingMap/red_shading.png) ![Green (even rows) lens shading map](android.statistics.lensShadingMap/green_e_shading.png) ![Green (odd rows) lens shading map](android.statistics.lensShadingMap/green_o_shading.png) ![Blue lens shading map](android.statistics.lensShadingMap/blue_shading.png) As a visualization only, inverting the full-color map to recover an image of a gray wall (using bicubic interpolation for visual quality) as captured by the sensor gives: ![Image of a uniform white wall (inverse shading map)](android.statistics.lensShadingMap/inv_shading.png) Note that the RAW image data might be subject to lens shading correction not reported on this map. Query android.sensor.info.lensShadingApplied to see if RAW image data has subject to lens shading correction. If android.sensor.info.lensShadingApplied is TRUE, the RAW image data is subject to partial or full lens shading correction. In the case full lens shading correction is applied to RAW images, the gain factor map reported in this key will contain all 1.0 gains. In other words, the map reported in this key is the remaining lens shading that needs to be applied on the RAW image to get images without lens shading artifacts. See android.request.maxNumOutputRaw for a list of RAW image formats.

The lens shading map calculation may depend on exposure and white balance statistics. When AE and AWB are in AUTO modes (android.control.aeMode `!=` OFF and android.control.awbMode `!=` OFF), the HAL may have all the information it need to generate most accurate lens shading map. When AE or AWB are in manual mode (android.control.aeMode `==` OFF or android.control.awbMode `==` OFF), the shading map may be adversely impacted by manual exposure or white balance parameters. To avoid generating unreliable shading map data, the HAL may choose to lock the shading map with the latest known good map generated when the AE and AWB are in AUTO modes. 4 The best-fit color channel gains calculated by the camera device's statistics units for the current output frame.

This may be different than the gains used for this frame, since statistics processing on data from a new frame typically completes after the transform has already been applied to that frame. The 4 channel gains are defined in Bayer domain, see android.colorCorrection.gains for details. This value should always be calculated by the auto-white balance (AWB) block, regardless of the android.control.* current values.

3 3 The best-fit color transform matrix estimate calculated by the camera device's statistics units for the current output frame.

The camera device will provide the estimate from its statistics unit on the white balance transforms to use for the next frame. These are the values the camera device believes are the best fit for the current output frame. This may be different than the transform used for this frame, since statistics processing on data from a new frame typically completes after the transform has already been applied to that frame. These estimates must be provided for all frames, even if capture settings and color transforms are set by the application. This value should always be calculated by the auto-white balance (AWB) block, regardless of the android.control.* current values.

NONE The camera device does not detect any flickering illumination in the current scene. 50HZ The camera device detects illumination flickering at 50Hz in the current scene. 60HZ The camera device detects illumination flickering at 60Hz in the current scene. The camera device estimated scene illumination lighting frequency.

Many light sources, such as most fluorescent lights, flicker at a rate that depends on the local utility power standards. This flicker must be accounted for by auto-exposure routines to avoid artifacts in captured images. The camera device uses this entry to tell the application what the scene illuminant frequency is. When manual exposure control is enabled (`android.control.aeMode == OFF` or `android.control.mode == OFF`), the android.control.aeAntibandingMode doesn't perform antibanding, and the application can ensure it selects exposure times that do not cause banding issues by looking into this metadata field. See android.control.aeAntibandingMode for more details. Reports NONE if there doesn't appear to be flickering illumination.

2 n List of `(x, y)` coordinates of hot/defective pixels on the sensor. n <= number of pixels on the sensor. The `(x, y)` coordinates must be bounded by android.sensor.info.pixelArraySize.

A coordinate `(x, y)` must lie between `(0, 0)`, and `(width - 1, height - 1)` (inclusive), which are the top-left and bottom-right of the pixel array, respectively. The width and height dimensions are given in android.sensor.info.pixelArraySize. This may include hot pixels that lie outside of the active array bounds given by android.sensor.info.activeArraySize.

A hotpixel map contains the coordinates of pixels on the camera sensor that do report valid values (usually due to defects in the camera sensor). This includes pixels that are stuck at certain values, or have a response that does not accuractly encode the incoming light from the scene. To avoid performance issues, there should be significantly fewer hot pixels than actual pixels on the camera sensor. OFF Do not include a lens shading map in the capture result. ON Include a lens shading map in the capture result. Whether the camera device will output the lens shading map in output result metadata. android.statistics.info.availableLensShadingMapModes

When set to ON, android.statistics.lensShadingMap will be provided in the output result metadata. ON is always supported on devices with the RAW capability.

n 2 Tonemapping / contrast / gamma curve for the blue channel, to use when android.tonemap.mode is CONTRAST_CURVE.

See android.tonemap.curveRed for more details.

n 2 Tonemapping / contrast / gamma curve for the green channel, to use when android.tonemap.mode is CONTRAST_CURVE.

See android.tonemap.curveRed for more details.

n 2 Tonemapping / contrast / gamma curve for the red channel, to use when android.tonemap.mode is CONTRAST_CURVE. 0-1 on both input and output coordinates, normalized as a floating-point value such that 0 == black and 1 == white.

Each channel's curve is defined by an array of control points: android.tonemap.curveRed = [ P0in, P0out, P1in, P1out, P2in, P2out, P3in, P3out, ..., PNin, PNout ] 2 <= N <= android.tonemap.maxCurvePoints These are sorted in order of increasing `Pin`; it is required that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points. Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints). A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness. Linear mapping: android.tonemap.curveRed = [ 0, 0, 1.0, 1.0 ] ![Linear mapping curve](android.tonemap.curveRed/linear_tonemap.png) Invert mapping: android.tonemap.curveRed = [ 0, 1.0, 1.0, 0 ] ![Inverting mapping curve](android.tonemap.curveRed/inverse_tonemap.png) Gamma 1/2.2 mapping, with 16 control points: android.tonemap.curveRed = [ 0.0000, 0.0000, 0.0667, 0.2920, 0.1333, 0.4002, 0.2000, 0.4812, 0.2667, 0.5484, 0.3333, 0.6069, 0.4000, 0.6594, 0.4667, 0.7072, 0.5333, 0.7515, 0.6000, 0.7928, 0.6667, 0.8317, 0.7333, 0.8685, 0.8000, 0.9035, 0.8667, 0.9370, 0.9333, 0.9691, 1.0000, 1.0000 ] ![Gamma = 1/2.2 tonemapping curve](android.tonemap.curveRed/gamma_tonemap.png) Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points: android.tonemap.curveRed = [ 0.0000, 0.0000, 0.0667, 0.2864, 0.1333, 0.4007, 0.2000, 0.4845, 0.2667, 0.5532, 0.3333, 0.6125, 0.4000, 0.6652, 0.4667, 0.7130, 0.5333, 0.7569, 0.6000, 0.7977, 0.6667, 0.8360, 0.7333, 0.8721, 0.8000, 0.9063, 0.8667, 0.9389, 0.9333, 0.9701, 1.0000, 1.0000 ] ![sRGB tonemapping curve](android.tonemap.curveRed/srgb_tonemap.png)

For good quality of mapping, at least 128 control points are preferred. A typical use case of this would be a gamma-1/2.2 curve, with as many control points used as are available. Tonemapping / contrast / gamma curve to use when android.tonemap.mode is CONTRAST_CURVE.

The tonemapCurve consist of three curves for each of red, green, and blue channels respectively. The following example uses the red channel as an example. The same logic applies to green and blue channel. Each channel's curve is defined by an array of control points: curveRed = [ P0(in, out), P1(in, out), P2(in, out), P3(in, out), ..., PN(in, out) ] 2 <= N <= android.tonemap.maxCurvePoints These are sorted in order of increasing `Pin`; it is always guaranteed that input values 0.0 and 1.0 are included in the list to define a complete mapping. For input values between control points, the camera device must linearly interpolate between the control points. Each curve can have an independent number of points, and the number of points can be less than max (that is, the request doesn't have to always provide a curve with number of points equivalent to android.tonemap.maxCurvePoints). A few examples, and their corresponding graphical mappings; these only specify the red channel and the precision is limited to 4 digits, for conciseness. Linear mapping: curveRed = [ (0, 0), (1.0, 1.0) ] ![Linear mapping curve](android.tonemap.curveRed/linear_tonemap.png) Invert mapping: curveRed = [ (0, 1.0), (1.0, 0) ] ![Inverting mapping curve](android.tonemap.curveRed/inverse_tonemap.png) Gamma 1/2.2 mapping, with 16 control points: curveRed = [ (0.0000, 0.0000), (0.0667, 0.2920), (0.1333, 0.4002), (0.2000, 0.4812), (0.2667, 0.5484), (0.3333, 0.6069), (0.4000, 0.6594), (0.4667, 0.7072), (0.5333, 0.7515), (0.6000, 0.7928), (0.6667, 0.8317), (0.7333, 0.8685), (0.8000, 0.9035), (0.8667, 0.9370), (0.9333, 0.9691), (1.0000, 1.0000) ] ![Gamma = 1/2.2 tonemapping curve](android.tonemap.curveRed/gamma_tonemap.png) Standard sRGB gamma mapping, per IEC 61966-2-1:1999, with 16 control points: curveRed = [ (0.0000, 0.0000), (0.0667, 0.2864), (0.1333, 0.4007), (0.2000, 0.4845), (0.2667, 0.5532), (0.3333, 0.6125), (0.4000, 0.6652), (0.4667, 0.7130), (0.5333, 0.7569), (0.6000, 0.7977), (0.6667, 0.8360), (0.7333, 0.8721), (0.8000, 0.9063), (0.8667, 0.9389), (0.9333, 0.9701), (1.0000, 1.0000) ] ![sRGB tonemapping curve](android.tonemap.curveRed/srgb_tonemap.png)

This entry is created by the framework from the curveRed, curveGreen and curveBlue entries. CONTRAST_CURVE Use the tone mapping curve specified in the android.tonemap.curve* entries. All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.curve. Must not slow down frame rate relative to raw sensor output. FAST Advanced gamma mapping and color enhancement may be applied, without reducing frame rate compared to raw sensor output. HIGH_QUALITY High-quality gamma mapping and color enhancement will be applied, at the cost of possibly reduced frame rate compared to raw sensor output. GAMMA_VALUE Use the gamma value specified in android.tonemap.gamma to peform tonemapping. All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.gamma. Must not slow down frame rate relative to raw sensor output. PRESET_CURVE Use the preset tonemapping curve specified in android.tonemap.presetCurve to peform tonemapping. All color enhancement and tonemapping must be disabled, except for applying the tonemapping curve specified by android.tonemap.presetCurve. Must not slow down frame rate relative to raw sensor output. High-level global contrast/gamma/tonemapping control. android.tonemap.availableToneMapModes

When switching to an application-defined contrast curve by setting android.tonemap.mode to CONTRAST_CURVE, the curve is defined per-channel with a set of `(in, out)` points that specify the mapping from input high-bit-depth pixel value to the output low-bit-depth value. Since the actual pixel ranges of both input and output may change depending on the camera pipeline, the values are specified by normalized floating-point numbers. More-complex color mapping operations such as 3D color look-up tables, selective chroma enhancement, or other non-linear color transforms will be disabled when android.tonemap.mode is CONTRAST_CURVE. When using either FAST or HIGH_QUALITY, the camera device will emit its own tonemap curve in android.tonemap.curve. These values are always available, and as close as possible to the actually used nonlinear/nonglobal transforms. If a request is sent with CONTRAST_CURVE with the camera device's provided curve in FAST or HIGH_QUALITY, the image's tonemap will be roughly the same.

Maximum number of supported points in the tonemap curve that can be used for android.tonemap.curve.

If the actual number of points provided by the application (in android.tonemap.curve*) is less than this maximum, the camera device will resample the curve to its internal representation, using linear interpolation. The output curves in the result metadata may have a different number of points than the input curves, and will represent the actual hardware curves used as closely as possible when linearly interpolated.

This value must be at least 64. This should be at least 128. n List of tonemapping modes for android.tonemap.mode that are supported by this camera device. Any value listed in android.tonemap.mode

Camera devices that support the MANUAL_POST_PROCESSING capability will always contain at least one of below mode combinations: * CONTRAST_CURVE, FAST and HIGH_QUALITY * GAMMA_VALUE, PRESET_CURVE, FAST and HIGH_QUALITY This includes all FULL level devices.

HAL must support both FAST and HIGH_QUALITY if automatic tonemap control is available on the camera device, but the underlying implementation can be the same for both modes. That is, if the highest quality implementation on the camera device does not slow down capture rate, then FAST and HIGH_QUALITY will generate the same output. Tonemapping curve to use when android.tonemap.mode is GAMMA_VALUE

The tonemap curve will be defined the following formula: * OUT = pow(IN, 1.0 / gamma) where IN and OUT is the input pixel value scaled to range [0.0, 1.0], pow is the power function and gamma is the gamma value specified by this key. The same curve will be applied to all color channels. The camera device may clip the input gamma value to its supported range. The actual applied value will be returned in capture result. The valid range of gamma value varies on different devices, but values within [1.0, 5.0] are guaranteed not to be clipped.

SRGB Tonemapping curve is defined by sRGB REC709 Tonemapping curve is defined by ITU-R BT.709 Tonemapping curve to use when android.tonemap.mode is PRESET_CURVE

The tonemap curve will be defined by specified standard. sRGB (approximated by 16 control points): ![sRGB tonemapping curve](android.tonemap.curveRed/srgb_tonemap.png) Rec. 709 (approximated by 16 control points): ![Rec. 709 tonemapping curve](android.tonemap.curveRed/rec709_tonemap.png) Note that above figures show a 16 control points approximation of preset curves. Camera devices may apply a different approximation to the curve.

OFF ON This LED is nominally used to indicate to the user that the camera is powered on and may be streaming images back to the Application Processor. In certain rare circumstances, the OS may disable this when video is processed locally and not transmitted to any untrusted applications. In particular, the LED *must* always be on when the data could be transmitted off the device. The LED *should* always be on whenever data is stored locally on the device. The LED *may* be off if a trusted application is using the data that doesn't violate the above rules. n TRANSMIT android.led.transmit control is used. A list of camera LEDs that are available on this system.

LIMITED This camera device has only limited capabilities. FULL This camera device is capable of supporting advanced imaging applications. LEGACY This camera device is running in backward compatibility mode. HIGH_RESOLUTION This camera device is capable of supporting advanced imaging applications at full rate, and additional high-resolution outputs at lower rates. Generally classifies the overall set of the camera device functionality.

Camera devices will come in three flavors: LEGACY, LIMITED and FULL. A FULL device will support below capabilities: * 30fps operation at maximum resolution (== sensor resolution) is preferred, more than 20fps is required, for at least uncompressed YUV output. (android.request.availableCapabilities contains BURST_CAPTURE) * Per frame control (android.sync.maxLatency `==` PER_FRAME_CONTROL) * Manual sensor control (android.request.availableCapabilities contains MANUAL_SENSOR) * Manual post-processing control (android.request.availableCapabilities contains MANUAL_POST_PROCESSING) * Arbitrary cropping region (android.scaler.croppingType `==` FREEFORM) * At least 3 processed (but not stalling) format output streams (android.request.maxNumOutputProc `>=` 3) * The required stream configuration defined in android.scaler.availableStreamConfigurations * The required exposure time range defined in android.sensor.info.exposureTimeRange * The required maxFrameDuration defined in android.sensor.info.maxFrameDuration A LIMITED device may have some or none of the above characteristics. To find out more refer to android.request.availableCapabilities. Some features are not part of any particular hardware level or capability and must be queried separately. These include: * Calibrated timestamps (android.sensor.info.timestampSource `==` REALTIME) * Precision lens control (android.lens.info.focusDistanceCalibration `==` CALIBRATED) * Face detection (android.statistics.info.availableFaceDetectModes) * Optical or electrical image stabilization (android.lens.info.availableOpticalStabilization, android.control.availableVideoStabilizationModes) A LEGACY device does not support per-frame control, manual sensor control, manual post-processing, arbitrary cropping regions, and has relaxed performance constraints. Each higher level supports everything the lower level supports in this order: FULL `>` LIMITED `>` LEGACY. A HIGH_RESOLUTION device is equivalent to a FULL device, except that: * At least one output resolution of 8 megapixels or higher in uncompressed YUV is supported at `>=` 20 fps. * Maximum-size (sensor resolution) uncompressed YUV is supported at `>=` 10 fps. * For devices that list the RAW capability and support either RAW10 or RAW12 output, maximum-resolution RAW10 or RAW12 capture will operate at least at the rate of maximum-resolution YUV capture, and at least one supported output resolution of 8 megapixels or higher in RAW10 or RAW12 is supported `>=` 20 fps.

The camera 3 HAL device can implement one of two possible operational modes; limited and full. Full support is expected from new higher-end devices. Limited mode has hardware requirements roughly in line with those for a camera HAL device v1 implementation, and is expected from older or inexpensive devices. Full is a strict superset of limited, and they share the same essential operational flow. For full details refer to "S3. Operational Modes" in camera3.h Camera HAL3+ must not implement LEGACY mode. It is there for backwards compatibility in the `android.hardware.camera2` user-facing API only.

OFF ON Whether black-level compensation is locked to its current values, or is free to vary.

When set to `true` (ON), the values used for black-level compensation will not change until the lock is set to `false` (OFF). Since changes to certain capture parameters (such as exposure time) may require resetting of black level compensation, the camera device must report whether setting the black level lock was successful in the output result metadata. For example, if a sequence of requests is as follows: * Request 1: Exposure = 10ms, Black level lock = OFF * Request 2: Exposure = 10ms, Black level lock = ON * Request 3: Exposure = 10ms, Black level lock = ON * Request 4: Exposure = 20ms, Black level lock = ON * Request 5: Exposure = 20ms, Black level lock = ON * Request 6: Exposure = 20ms, Black level lock = ON And the exposure change in Request 4 requires the camera device to reset the black level offsets, then the output result metadata is expected to be: * Result 1: Exposure = 10ms, Black level lock = OFF * Result 2: Exposure = 10ms, Black level lock = ON * Result 3: Exposure = 10ms, Black level lock = ON * Result 4: Exposure = 20ms, Black level lock = OFF * Result 5: Exposure = 20ms, Black level lock = ON * Result 6: Exposure = 20ms, Black level lock = ON This indicates to the application that on frame 4, black levels were reset due to exposure value changes, and pixel values may not be consistent across captures. The camera device will maintain the lock to the extent possible, only overriding the lock to OFF when changes to other request parameters require a black level recalculation or reset.

If for some reason black level locking is no longer possible (for example, the analog gain has changed, which forces black level offsets to be recalculated), then the HAL must override this request (and it must report 'OFF' when this does happen) until the next capture for which locking is possible again.

Whether the black level offset was locked for this frame. Should be ON if android.blackLevel.lock was ON in the capture request, unless a change in other capture settings forced the camera device to perform a black level reset.

CONVERGING The current result is not yet fully synchronized to any request. Synchronization is in progress, and reading metadata from this result may include a mix of data that have taken effect since the last synchronization time. In some future result, within android.sync.maxLatency frames, this value will update to the actual frame number frame number the result is guaranteed to be synchronized to (as long as the request settings remain constant). UNKNOWN The current result's synchronization status is unknown. The result may have already converged, or it may be in progress. Reading from this result may include some mix of settings from past requests. After a settings change, the new settings will eventually all take effect for the output buffers and results. However, this value will not change when that happens. Altering settings rapidly may provide outcomes using mixes of settings from recent requests. This value is intended primarily for backwards compatibility with the older camera implementations (for android.hardware.Camera). The frame number corresponding to the last request with which the output result (metadata + buffers) has been fully synchronized. Either a non-negative value corresponding to a `frame_number`, or one of the two enums (CONVERGING / UNKNOWN).

When a request is submitted to the camera device, there is usually a delay of several frames before the controls get applied. A camera device may either choose to account for this delay by implementing a pipeline and carefully submit well-timed atomic control updates, or it may start streaming control changes that span over several frame boundaries. In the latter case, whenever a request's settings change relative to the previous submitted request, the full set of changes may take multiple frame durations to fully take effect. Some settings may take effect sooner (in less frame durations) than others. While a set of control changes are being propagated, this value will be CONVERGING. Once it is fully known that a set of control changes have been finished propagating, and the resulting updated control settings have been read back by the camera device, this value will be set to a non-negative frame number (corresponding to the request to which the results have synchronized to). Older camera device implementations may not have a way to detect when all camera controls have been applied, and will always set this value to UNKNOWN. FULL capability devices will always have this value set to the frame number of the request corresponding to this result. _Further details_: * Whenever a request differs from the last request, any future results not yet returned may have this value set to CONVERGING (this could include any in-progress captures not yet returned by the camera device, for more details see pipeline considerations below). * Submitting a series of multiple requests that differ from the previous request (e.g. r1, r2, r3 s.t. r1 != r2 != r3) moves the new synchronization frame to the last non-repeating request (using the smallest frame number from the contiguous list of repeating requests). * Submitting the same request repeatedly will not change this value to CONVERGING, if it was already a non-negative value. * When this value changes to non-negative, that means that all of the metadata controls from the request have been applied, all of the metadata controls from the camera device have been read to the updated values (into the result), and all of the graphics buffers corresponding to this result are also synchronized to the request. _Pipeline considerations_: Submitting a request with updated controls relative to the previously submitted requests may also invalidate the synchronization state of all the results corresponding to currently in-flight requests. In other words, results for this current request and up to android.request.pipelineMaxDepth prior requests may have their android.sync.frameNumber change to CONVERGING.

Using UNKNOWN here is illegal unless android.sync.maxLatency is also UNKNOWN. FULL capability devices should simply set this value to the `frame_number` of the request this result corresponds to. PER_FRAME_CONTROL Every frame has the requests immediately applied. Changing controls over multiple requests one after another will produce results that have those controls applied atomically each frame. All FULL capability devices will have this as their maxLatency. UNKNOWN Each new frame has some subset (potentially the entire set) of the past requests applied to the camera settings. By submitting a series of identical requests, the camera device will eventually have the camera settings applied, but it is unknown when that exact point will be. All LEGACY capability devices will have this as their maxLatency. The maximum number of frames that can occur after a request (different than the previous) has been submitted, and before the result's state becomes synchronized. Frame counts A positive value, PER_FRAME_CONTROL, or UNKNOWN.

This defines the maximum distance (in number of metadata results), between the frame number of the request that has new controls to apply and the frame number of the result that has all the controls applied. In other words this acts as an upper boundary for how many frames must occur before the camera device knows for a fact that the new submitted camera settings have been applied in outgoing frames.

For example if maxLatency was 2, initial request = X (repeating) request1 = X request2 = Y request3 = Y request4 = Y where requestN has frameNumber N, and the first of the repeating initial request's has frameNumber F (and F < 1). initial result = X' + { android.sync.frameNumber == F } result1 = X' + { android.sync.frameNumber == F } result2 = X' + { android.sync.frameNumber == CONVERGING } result3 = X' + { android.sync.frameNumber == CONVERGING } result4 = X' + { android.sync.frameNumber == 2 } where resultN has frameNumber N. Since `result4` has a `frameNumber == 4` and `android.sync.frameNumber == 2`, the distance is clearly `4 - 2 = 2`. Use `frame_count` from camera3_request_t instead of android.request.frameCount or `@link{android.hardware.camera2.CaptureResult#getFrameNumber}`. LIMITED devices are strongly encouraged to use a non-negative value. If UNKNOWN is used here then app developers do not have a way to know when sensor settings have been applied.

The amount of exposure time increase factor applied to the original output frame by the application processing before sending for reprocessing. Relative exposure time increase factor. >= 1.0

This is optional, and will be supported if the camera device supports YUV_REPROCESSING capability (android.request.availableCapabilities contains YUV_REPROCESSING). For some YUV reprocessing use cases, the application may choose to filter the original output frames to effectively reduce the noise to the same level as a frame that was captured with longer exposure time. To be more specific, assuming the original captured images were captured with a sensitivity of S and an exposure time of T, the model in the camera device is that the amount of noise in the image would be approximately what would be expected if the original capture parameters had been a sensitivity of S/effectiveExposureFactor and an exposure time of T*effectiveExposureFactor, rather than S and T respectively. If the captured images were processed by the application before being sent for reprocessing, then the application may have used image processing algorithms and/or multi-frame image fusion to reduce the noise in the application-processed images (input images). By using the effectiveExposureFactor control, the application can communicate to the camera device the actual noise level improvement in the application-processed image. With this information, the camera device can select appropriate noise reduction and edge enhancement parameters to avoid excessive noise reduction (android.noiseReduction.mode) and insufficient edge enhancement (android.edge.mode) being applied to the reprocessed frames. For example, for multi-frame image fusion use case, the application may fuse multiple output frames together to a final frame for reprocessing. When N image are fused into 1 image for reprocessing, the exposure time increase factor could be up to square root of N (based on a simple photon shot noise model). The camera device will adjust the reprocessing noise reduction and edge enhancement parameters accordingly to produce the best quality images. This is relative factor, 1.0 indicates the application hasn't processed the input buffer in a way that affects its effective exposure time. This control is only effective for YUV reprocessing capture request. For noise reduction reprocessing, it is only effective when `android.noiseReduction.mode != OFF`. Similarly, for edge enhancement reprocessing, it is only effective when `android.edge.mode != OFF`.

The maximal camera capture pipeline stall (in unit of frame count) introduced by a reprocess capture request. Number of frames. <= 4

The key describes the maximal interference that one reprocess (input) request can introduce to the camera simultaneous streaming of regular (output) capture requests, including repeating requests. When a reprocessing capture request is submitted while a camera output repeating request (e.g. preview) is being served by the camera device, it may preempt the camera capture pipeline for at least one frame duration so that the camera device is unable to process the following capture request in time for the next sensor start of exposure boundary. When this happens, the application may observe a capture time gap (longer than one frame duration) between adjacent capture output frames, which usually exhibits as preview glitch if the repeating request output targets include a preview surface. This key gives the worst-case number of frame stall introduced by one reprocess request with any kind of formats/sizes combination. If this key reports 0, it means a reprocess request doesn't introduce any glitch to the ongoing camera repeating request outputs, as if this reprocess request is never issued. This key is supported if the camera device supports PRIVATE or YUV reprocessing ( i.e. android.request.availableCapabilities contains PRIVATE_REPROCESSING or YUV_REPROCESSING).

Maximum number of points that a depth point cloud may contain.

If a camera device supports outputting depth range data in the form of a depth point cloud ({@link android.graphics.ImageFormat#DEPTH_POINT_CLOUD}), this is the maximum number of points an output buffer may contain. Any given buffer may contain between 0 and maxDepthSamples points, inclusive. If output in the depth point cloud format is not supported, this entry will not be defined.

n 4 OUTPUT INPUT The available depth dataspace stream configurations that this camera device supports (i.e. format, width, height, output/input stream).

These are output stream configurations for use with dataSpace HAL_DATASPACE_DEPTH. The configurations are listed as `(format, width, height, input?)` tuples. Only devices that support depth output for at least the HAL_PIXEL_FORMAT_Y16 dense depth map may include this entry. A device that also supports the HAL_PIXEL_FORMAT_BLOB sparse depth point cloud must report a single entry for the format in this list as `(HAL_PIXEL_FORMAT_BLOB, android.depth.maxDepthSamples, 1, OUTPUT)` in addition to the entries for HAL_PIXEL_FORMAT_Y16.

4 n This lists the minimum frame duration for each format/size combination for depth output formats. (format, width, height, ns) x n

4 n This lists the maximum stall duration for each output format/size combination for depth streams. (format, width, height, ns) x n

A stall duration is how much extra time would get added to the normal minimum frame duration for a repeating request that has streams with non-zero stall. This functions similarly to android.scaler.availableStallDurations for depth streams. All depth output stream formats may have a nonzero stall duration.

FALSE TRUE Indicates whether a capture request may target both a DEPTH16 / DEPTH_POINT_CLOUD output, and normal color outputs (such as YUV_420_888, JPEG, or RAW) simultaneously.

If TRUE, including both depth and color outputs in a single capture request is not supported. An application must interleave color and depth requests. If FALSE, a single request can target both types of output. Typically, this restriction exists on camera devices that need to emit a specific pattern or wavelength of light to measure depth values, which causes the color image to be corrupted during depth measurement.