Merge "media: Add MediaCodec constants for RoI encoding" into main (d821064a) · Commits · e / os / android_frameworks_base

core/api/current.txt

+3 −0

media/java/android/media/MediaCodec.java

+66 −2

Original line number	Original line	Diff line number	Diff line
	@@ -16,6 +16,10 @@

	package android.media;		package android.media;

			import static android.media.codec.Flags.FLAG_REGION_OF_INTEREST;

			import static com.android.media.codec.flags.Flags.FLAG_LARGE_AUDIO_FRAME;

	import android.Manifest;		import android.Manifest;
	import android.annotation.FlaggedApi;		import android.annotation.FlaggedApi;
	import android.annotation.IntDef;		import android.annotation.IntDef;
	@@ -51,7 +55,6 @@ import java.util.BitSet;
	import java.util.Collections;		import java.util.Collections;
	import java.util.HashMap;		import java.util.HashMap;
	import java.util.HashSet;		import java.util.HashSet;
	import java.util.Iterator;
	import java.util.List;		import java.util.List;
	import java.util.Map;		import java.util.Map;
	import java.util.Objects;		import java.util.Objects;
	@@ -62,7 +65,6 @@ import java.util.concurrent.LinkedBlockingQueue;
	import java.util.concurrent.locks.Lock;		import java.util.concurrent.locks.Lock;
	import java.util.concurrent.locks.ReentrantLock;		import java.util.concurrent.locks.ReentrantLock;

	import static com.android.media.codec.flags.Flags.FLAG_LARGE_AUDIO_FRAME;
	/**		/**
	MediaCodec class can be used to access low-level media codecs, i.e. encoder/decoder components.		MediaCodec class can be used to access low-level media codecs, i.e. encoder/decoder components.
	It is part of the Android low-level multimedia support infrastructure (normally used together		It is part of the Android low-level multimedia support infrastructure (normally used together
	@@ -4935,6 +4937,68 @@ final public class MediaCodec {
	*/		*/
	public static final String PARAMETER_KEY_TUNNEL_PEEK = "tunnel-peek";		public static final String PARAMETER_KEY_TUNNEL_PEEK = "tunnel-peek";

			/**
			* Set the region of interest as QpOffset-Map on the next queued input frame.
			* <p>
			* The associated value is a byte array containing quantization parameter (QP) offsets in
			* raster scan order for the entire frame at 16x16 granularity. The size of the byte array
			* shall be ((frame_width + 15) / 16) * ((frame_height + 15) / 16), where frame_width and
			* frame_height correspond to width and height configured using {@link MediaFormat#KEY_WIDTH}
			* and {@link MediaFormat#KEY_HEIGHT} keys respectively. During encoding, if the coding unit
			* size is larger than 16x16, then the qpOffset information of all 16x16 blocks that
			* encompass the coding unit is combined and used. The QP of target block will be calculated
			* as 'frameQP + offsetQP'. If the result exceeds minQP or maxQP configured then the value
			* may be clamped. Negative offset results in blocks encoded at lower QP than frame QP and
			* positive offsets will result in encoding blocks at higher QP than frame QP. If the areas
			* of negative QP and positive QP are chosen wisely, the overall viewing experience can be
			* improved.
			* <p>
			* If byte array size is too small than the expected size, components may ignore the
			* configuration silently. If the byte array exceeds the expected size, components shall use
			* the initial portion and ignore the rest.
			* <p>
			* The scope of this key is throughout the encoding session until it is reconfigured during
			* running state.
			* <p>
			* @see #setParameters(Bundle)
			*/
			@FlaggedApi(FLAG_REGION_OF_INTEREST)
			public static final String PARAMETER_KEY_QP_OFFSET_MAP = "qp-offset-map";

			/**
			* Set the region of interest as QpOffset-Rects on the next queued input frame.
			* <p>
			* The associated value is a String in the format "Top1,Left1-Bottom1,Right1=Offset1;Top2,
			* Left2-Bottom2,Right2=Offset2;...". Co-ordinates (Top, Left), (Top, Right), (Bottom, Left)
			* and (Bottom, Right) form the vertices of bounding box of region of interest in pixels.
			* Pixel (0, 0) points to the top-left corner of the frame. Offset is the suggested
			* quantization parameter (QP) offset of the blocks in the bounding box. The bounding box
			* will get stretched outwards to align to LCU boundaries during encoding. The Qp Offset is
			* integral and shall be in the range [-128, 127]. The QP of target block will be calculated
			* as frameQP + offsetQP. If the result exceeds minQP or maxQP configured then the value may
			* be clamped. Negative offset results in blocks encoded at lower QP than frame QP and
			* positive offsets will result in blocks encoded at higher QP than frame QP. If the areas of
			* negative QP and positive QP are chosen wisely, the overall viewing experience can be
			* improved.
			* <p>
			* If Roi rect is not valid that is bounding box width is < 0 or bounding box height is < 0,
			* components may ignore the configuration silently. If Roi rect extends outside frame
			* boundaries, then rect shall be clamped to the frame boundaries.
			* <p>
			* The scope of this key is throughout the encoding session until it is reconfigured during
			* running state.
			* <p>
			* The maximum number of contours (rectangles) that can be specified for a given input frame
			* is device specific. Implementations will drop/ignore the rectangles that are beyond their
			* supported limit. Hence it is preferable to place the rects in descending order of
			* importance. Transitively, if the bounding boxes overlap, then the most preferred
			* rectangle's qp offset (earlier rectangle qp offset) will be used to quantize the block.
			* <p>
			* @see #setParameters(Bundle)
			*/
			@FlaggedApi(FLAG_REGION_OF_INTEREST)
			public static final String PARAMETER_KEY_QP_OFFSET_RECTS = "qp-offset-rects";

	/**		/**
	* Communicate additional parameter changes to the component instance.		* Communicate additional parameter changes to the component instance.
	* <b>Note:</b> Some of these parameter changes may silently fail to apply.		* <b>Note:</b> Some of these parameter changes may silently fail to apply.

media/java/android/media/MediaCodecInfo.java

+40 −0

Original line number	Original line	Diff line number	Diff line
	@@ -20,6 +20,7 @@ import static android.media.Utils.intersectSortedDistinctRanges;
	import static android.media.Utils.sortDistinctRanges;		import static android.media.Utils.sortDistinctRanges;
	import static android.media.codec.Flags.FLAG_DYNAMIC_COLOR_ASPECTS;		import static android.media.codec.Flags.FLAG_DYNAMIC_COLOR_ASPECTS;
	import static android.media.codec.Flags.FLAG_HLG_EDITING;		import static android.media.codec.Flags.FLAG_HLG_EDITING;
			import static android.media.codec.Flags.FLAG_REGION_OF_INTEREST;

	import android.annotation.FlaggedApi;		import android.annotation.FlaggedApi;
	import android.annotation.IntRange;		import android.annotation.IntRange;
	@@ -740,6 +741,42 @@ public final class MediaCodecInfo {
	@FlaggedApi(FLAG_DYNAMIC_COLOR_ASPECTS)		@FlaggedApi(FLAG_DYNAMIC_COLOR_ASPECTS)
	public static final String FEATURE_DynamicColorAspects = "dynamic-color-aspects";		public static final String FEATURE_DynamicColorAspects = "dynamic-color-aspects";

			/**
			* <b>video encoder only</b>: codec supports region of interest encoding.
			* <p>
			* RoI encoding support means the codec accepts information that specifies the relative
			* importance of different portions of each video frame. This allows the encoder to
			* separate a video frame into critical and non-critical regions, and use more bits
			* (better quality) to represent the critical regions and de-prioritize non-critical
			* regions. In other words, the encoder chooses a negative qp bias for the critical
			* portions and a zero or positive qp bias for the non-critical portions.
			* <p>
			* At a basic level, if the encoder decides to encode each frame with a uniform
			* quantization value 'qpFrame' and a 'qpBias' is chosen/suggested for an LCU of the
			* frame, then the actual qp of the LCU will be 'qpFrame + qpBias', although this value
			* can be clamped basing on the min-max configured qp bounds for the current encoding
			* session.
			* <p>
			* In a shot, if a group of LCUs pan out quickly they can be marked as non-critical
			* thereby enabling the encoder to reserve fewer bits during their encoding. Contrarily,
			* LCUs that remain in shot for a prolonged duration can be encoded at better quality in
			* one frame thereby setting-up an excellent long-term reference for all future frames.
			* <p>
			* Note that by offsetting the quantization of each LCU, the overall bit allocation will
			* differ from the originally estimated bit allocation, and the encoder will adjust the
			* frame quantization for subsequent frames to meet the bitrate target. An effective
			* selection of critical regions can set-up a golden reference and this can compensate
			* for the bit burden that was introduced due to encoding RoI's at better quality.
			* On the other hand, an ineffective choice of critical regions might increase the
			* quality of certain parts of the image but this can hamper quality in subsequent frames.
			* <p>
			* @see MediaCodec#PARAMETER_KEY_QP_OFFSET_MAP
			* @see MediaCodec#PARAMETER_KEY_QP_OFFSET_RECTS
			*/
			@SuppressLint("AllUpper")
			@FlaggedApi(FLAG_REGION_OF_INTEREST)
			public static final String FEATURE_Roi = "region-of-interest";

	/**		/**
	* Query codec feature capabilities.		* Query codec feature capabilities.
	* <p>		* <p>
	@@ -798,6 +835,9 @@ public final class MediaCodecInfo {
	if (android.media.codec.Flags.hlgEditing()) {		if (android.media.codec.Flags.hlgEditing()) {
	features.add(new Feature(FEATURE_HlgEditing, (1 << 6), true));		features.add(new Feature(FEATURE_HlgEditing, (1 << 6), true));
	}		}
			if (android.media.codec.Flags.regionOfInterest()) {
			features.add(new Feature(FEATURE_Roi, (1 << 7), true));
			}

	// feature to exclude codec from REGULAR codec list		// feature to exclude codec from REGULAR codec list
	features.add(new Feature(FEATURE_SpecialCodec, (1 << 30), false, true));		features.add(new Feature(FEATURE_SpecialCodec, (1 << 30), false, true));