This test was testing a scenario that never happens in production.
Indeed, in this test, a frame was registered to the
ExternalTextureManager but never rendered on the decoder ouput surface.
This never happens in production because both actions are normally
performed in
PlaybackVideoGraphWrapper.InputVideoSink.handleInputFrame().
This test was also failing on API 31 emulator. As the frame was never
rendered, the timeout to release all registered frames in
ExternalTextureManager was always exceeded. This sometimes caused the
decoder to fail on this emulator because there was no interaction with
the decoder for a long time.
PiperOrigin-RevId: 705490910
This is inspired from TestExoPlayerBuilder.
Adding this class has a few advantages:
- It makes testing easier for apps by configuring Transformer for unit
tests (for example, by setting the clock).
- It removes the Transformer setters that apps should not use for unit
testing (for example, the video MIME type should not be set because
it would cause Transformer to transcode.
- It allows us to add additional setters in the future, for example to
build a Transformer that always fails.
PiperOrigin-RevId: 704181927
Add FrameExtraction.setMediaItem() method
Prefer non-vendor software codecs by default because
vendor codecs sometimes fail when we attempt reuse.
PiperOrigin-RevId: 703486235
After removing `Transformer.Builder.setFlattenForSlowMotion()`, there is no need to keep `flattenForSlowMotion` in Transformer since it's now set in `EditedMediaItem`.
The change also includes making `audioProcessors` and `videoEffects` attributes final.
PiperOrigin-RevId: 703076023
* Select an encoder that supports HDR editing.
* Set KEY_PROFILE to an HDR10 option
* Use DecodeOneFrameUtil test util to return the MediaCodec format,
which includes HDR_STATIC_INFO
PiperOrigin-RevId: 702752639
Earlier this year, we explored frame extraction built on top of Transformer
or ImageReader.
We decided against these approaches because random-access of frames
is a key requirement. And ImageReader behaviour is unreliable.
PiperOrigin-RevId: 702303490
This method was taking positionUs and elapsedRealtimeUs parameters and
throwing a VideoSinkException because it was calling render() to make
room for a new input frame. Instead of doing that, this CL makes sure
render() is called before handleInputFrame() in the renderer (even
though it may not make a significant difference).
PiperOrigin-RevId: 702273587
Previously cancelling one FrameExtractor.getFrame
ListenableFuture caused the following requests to fail.
Implement ability to cancel Futures, and correctly issue
next player seek commands after all past requests complete.
Removes uses of SettableFuture, and replaces them with
CallbackToFutureAdapter.
Adds a dependency on androidx.concurrent:concurrent-futures
PiperOrigin-RevId: 701941403
Frame Extractor was getting stuck with SeekParameters.CLOSEST_SYNC.
onPositionDiscontinuity callback was sometimes being called with a
non-adjusted new position.
Fix this by monitoring player state ready.
For the player to become ready, we have to override renderer isReady.
PiperOrigin-RevId: 701924752
The isLastBuffer flag passed to MediaCodecRenderer.processOutputBuffer()
is unreliable. It is true for the last frames (plural) of a video
stream, which makes it possible for the video renderer to end before all
the frames have been rendered to the output surface.
PiperOrigin-RevId: 700941685
Introduced a `mediaSourceFactory` attribute to CompositionPlayer. This allows for the use of a custom MediaSource.Factory when creating media sources, providing more flexibility and control over media source creation. Previously, the default `MediaSource.Factory` was used in all cases.
PiperOrigin-RevId: 700391911
Support extracting frames from HDR input by tone mapping
to SDR (BT.709).
ExperimentalFrameExtractor must be public because HDR tests
live in a different package.
PiperOrigin-RevId: 699994112
Adds a configuration option to Frame Extractor to choose MediaCodecSelector.
Add a MediaCodecSelector that lists software decoders first
PiperOrigin-RevId: 699962365
Renderers join when they can start processing but not produce output.
In CompositionPlayer before this change, the VideoSink will be flushed when
position is reset regardless. This is useful for seeking, as seeking triggers position reset. But with pre-warming, a video renderer can be joining - it is
enabled (which also triggers position reset) before the previous image renderer
is disabled. At this moment, as the image renderer is still producing frames,
we should not flush the video sink just now.
PiperOrigin-RevId: 699943882
Changes includes making sure the same `defaultMediaSourceFactory` is used when creating a new media source in `setPrimaryPlayerSequence` and `setSecondaryPlayerSequence`.
This is pre-work for introducing a new attribute `mediaSourceFactory` that can be optionally used instead of the default one. This will allow developers to pass a customized media source factory to CompositionPlayer.
PiperOrigin-RevId: 698721139
Adds a Frame extractor test that verifies decoder
respects rotation metadata from the mp4 container.
Do not rely on the MediaCodec decoder rotate the input.
Rotate via a video effect instead.
PiperOrigin-RevId: 698381282
3P app reported it's slow to see the first image frame on the screen, when
`playWhenReady` is false. This is because the player would wake up less
frequently when `playWhenReady` is false. This CL adds a runnable to wake up
the player any time a new frame is made available.
PiperOrigin-RevId: 698066887
Earlier watchdog timer was started only after all the tracks are
added. For the cases where export is stuck before adding tracks,
the export would hang forever and will not timeout automatically.
With the changes, watchdog timer is started as soon as export is
started/resumed. Now if the pipeline is stuck before writing any samples,
it will timeout and report error to the caller.
Existing test case: c35a9d62ba/libraries/transformer/src/test/java/androidx/media3/transformer/MediaItemExportTest.java (L996)
PiperOrigin-RevId: 695781654
A skeleton implementation that doesn't actually return decoded frames.
In the current state, we use ExoPlayer to seek to a position, and
ExoPlayer.setVideoEffects to record the presentation time selected.
Seeking and processing frames are synchronized via ListenableFuture
callbacks.
PiperOrigin-RevId: 695691183
In some cases, callers will need to pass in both BitmapLoader and Codec.DecoderFactory without specifying a custom MediaSource.Factory. Omitting the annotation will result in NULL_FOR_NONNULL_TYPE compilation errors in Kotlin.
PiperOrigin-RevId: 695481606
Enable transformer to transmux into alternative sample MIME types.
For example, some Dolby Vision profiles have a backwards-compatible
AVC or HEVC layer. MV-HEVC is backwards compatible with HEVC.
This change enables Transformer to transmux into the backwards compatible
format to improve compatibility with legacy APIs such as
MediaMetadataRetriever.
PiperOrigin-RevId: 693667597
Also makes muxer shift the first video timestamp to zero, if it's not.
The trim position should respect the media timeline.
For example in a video that is 10s long (without edit list), if an edit list
adds 1_000ms to each video sample, and trimming 100ms, here's the expected:
- The video duration is 10.9s (`10s + 1s edit - 0.1s trim`)
- The first video frame time would be at 0.9s (`1s edit - 0.1s trim`)
PiperOrigin-RevId: 692187399
