Stream Configurations

Android 10 introduces features allowing camera clients to choose optimal camera streams for specific use cases and to ensure that certain stream combinations are supported by the camera device. A stream configuration refers to a single camera stream configured in the camera device and a stream combination refers to one or more sets of streams configured in the camera device. For more on these features, see recommended stream configurations and API to query stream combinations.

Reference implementation

There is a vendor-side reference implementation of the recommended configuration streams and the API to query stream combination features. You can find this implementation at QCamera3HWI.cpp

Camera vendors can advertise recommended stream configurations for specific use cases to camera clients. These recommended stream configurations, which are subsets of StreamConfigurationMap, can help camera clients choose optimal configurations.

Although StreamConfigurationMap provides exhaustive stream configuration information to camera clients, it doesn't provide any information about the efficiency, power, or performance impacts of choosing one stream over another. Camera clients can freely choose from all the possible stream configurations but in many cases, this leads to clients using sub-optimal camera configurations and applications making time-consuming exhaustive searches.

For example, although some processed YUV formats are required and must be supported, the camera device might not have native support for the formats. This results in an additional processing pass for the format conversion and reduces efficiency. The size and corresponding aspect ratio can also have a similar impact making particular dimensions preferable in terms of power and performance.

Your recommended stream configuration maps aren't required to be exhaustive compared to the StreamConfigurationMap. The suggested configuration maps must follow the requirements in the implementation section and can include any of the available formats, sizes, or other values found in StreamConfigurationMap. Hidden formats, sizes, or other values not found in StreamConfigurationMap can't be included in recommended stream configuration maps.

All tests remain unchanged and aren't relaxed depending on the recommended stream configurations.

The recommended stream configurations provided by the camera implementation are optional and the camera client can ignore them.


Follow these steps to implement this feature.

Metadata entries

To enable this feature the Camera HAL must populate the following static metadata entries:

  • android.scaler.availableRecommendedStreamConfigurations: The recommended subset of stream configurations for specific use cases. The declaration uses simple bitmaps indicating the suggested use cases in the form of [1 << PREVIEW | 1 << RECORD..]. The use cases extend the regular (format, width, height, input) tuple with one additional entry. Non-existing public use cases or any other bits set within the range [PUBLIC_END, VENDOR_START] are prohibited.

    This information is stored in the availableRecommendedStreamConfigurations metadata tag.

    The following example shows an array for a recommended stream configuration for a camera device that only supports 4K and 1080p, where both resolutions are preferred for video recording but only 1080p is suggested for preview.

  • android.depth.availableRecommendedDepthStreamConfigurations (available only if supported by device): The recommended depth dataspace stream configurations suggested for this camera device. Similar to the above metadata entry, an additional use case bitmap indicates the suggested use cases.

    This information is stored in the availableRecommendedInputOutputFormatsMap metadata tag.

  • android.scaler.availableRecommendedInputOutputFormatsMap (available only if supported by device): The mapping of recommended image formats that are suggested for this camera device for input streams, to their corresponding output formats.

    This information is stored in the availableRecommendedDepthStreamConfigurations metadata tag.

This information is available to camera clients through the RecommendedStreamConfigurationMap API.

Required use cases

Recommended stream configurations must be provided for the following use cases and meet the corresponding requirements:

Use case Requirement
PREVIEW A preview must only include non-stalling processed stream configurations with output formats such as YUV_420_888 and IMPLEMENTATION_DEFINED.
RECORD A video record must include stream configurations that match the advertised supported media profiles with the IMPLEMENTATION_DEFINED format.
VIDEO_SNAPSHOT A video snapshot must include stream configurations that are at least as large as the maximum RECORD resolutions and only with the BLOB + DATASPACE_JFIF format/dataspace combination (JPEG). The configurations should not cause preview glitches and should be able to run at 30 fps.
SNAPSHOT Snapshot stream configurations must include at least one with a size close to with the BLOB + DATASPACE_JFIF format/dataspace combination (JPEG). Taking into account restrictions on aspect ratio, alignment, and other vendor-specific restrictions, the area of the maximum suggested size should not be less than 97% of the sensor array size area.
ZSL (if supported) If supported by the camera device, recommended input stream configurations must only be advertised together with other processed or stalling output formats.
RAW (if supported) If supported by the camera device, recommended raw stream configurations must only include RAW based output formats.

Other use cases

You can provide additional recommended configuration streams for use cases specific to your implementation.


To test your implementation of the recommended configuration streams, run the following CTS and VTS tests:

API to query stream combinations

The Android platform supports an API to query stream combinations. Implementing this API allows camera clients to safely query stream combinations at any point after receiving a valid CameraDevice instance, removing the overhead of initializing a camera capture session and the potential of having subsequent camera exceptions including camera breakage, and allowing for faster queries.

This feature also allows camera clients to receive a list of stream combinations compiled according to the guidelines for CameraDevice and the supported HW level. CTS tests are available to enforce the correctness of query results as much as possible covering a minor subset of the most common stream combinations.

You can choose to support this feature by implementing one additional HIDL API call in the Camera HAL.


To support an API to query stream combinations, the Camera HAL must provide an implementation for the isStreamCombinationSupported HIDL API interface. This interface checks whether the camera device supports a specified camera stream combination.

When called, the API must return one of the following status codes:

  • OK: The stream combination query was successful.
  • METHOD_NOT_SUPPORTED: The camera device does not support the stream combination query.
  • INTERNAL_ERROR: The stream combination query cannot complete due to an internal error.

The API returns true if the stream combination is supported. Otherwise, it returns false.

The framework uses the public API isSessionConfigurationSupported to check whether the particular session configuration is supported by the camera device.

Calls to the API must not have any side effects on normal camera operation. API calls must not alter any internal states or slow down the camera performance. Make sure that after the Camera HAL successfully validates a stream combination, camera clients can successfully configure the stream combination. To avoid issues, make sure the implementation does not store any information during stream combination queries, change its internal state, or engage in time-consuming operations.


To validate this feature, run the following camera CTS and VTS test cases:

Camera CTS modules:

Camera VTS: