Merge pull request #1055 from AradiPatrik:z-order-fix

PiperOrigin-RevId: 631616905

RELEASENOTES.md

@@ -62,6 +62,8 @@
     *   Change default SDR color working space from linear colors to electrical
         BT 709 SDR video. Also provides third option to retain the original
         colorspace.
+    *   Allow defining indeterminate z-order of EditedMediaItemSequences
+        ([#1055](https://github.com/androidx/media/pull/1055)).
 *   Muxers:
 *   IMA extension:
     *   Promote API that is required for apps to play

libraries/common/src/main/java/androidx/media3/common/VideoGraph.java

@@ -16,6 +16,7 @@
 
 package androidx.media3.common;
 
+import androidx.annotation.IntRange;
 import androidx.annotation.Nullable;
 import androidx.media3.common.util.UnstableApi;
 
@@ -73,19 +74,22 @@ public interface VideoGraph {
    * <p>A underlying processing {@link VideoFrameProcessor} is created every time this method is
    * called.
    *
+   * <p>All inputs must be registered before rendering frames to the underlying {@link
+   * #getProcessor(int) VideoFrameProcessor}.
+   *
    * <p>If the method throws, the caller must call {@link #release}.
    *
-   * @return The id of the registered input, which can be used to get the underlying {@link
+   * @param inputIndex The index of the input which could be used to order the inputs. The index
-   *     VideoFrameProcessor} via {@link #getProcessor(int)}.
+   *     must start from 0.
    */
-  int registerInput() throws VideoFrameProcessingException;
+  void registerInput(@IntRange(from = 0) int inputIndex) throws VideoFrameProcessingException;
 
   /**
    * Returns the {@link VideoFrameProcessor} that handles the processing for an input registered via
-   * {@link #registerInput()}. If the {@code inputId} is not {@linkplain #registerInput()
+   * {@link #registerInput(int)}. If the {@code inputIndex} is not {@linkplain #registerInput(int)
    * registered} before, this method will throw an {@link IllegalStateException}.
    */
-  VideoFrameProcessor getProcessor(int inputId);
+  VideoFrameProcessor getProcessor(int inputIndex);
 
   /**
    * Sets the output surface and supporting information.

libraries/effect/src/main/java/androidx/media3/effect/DefaultVideoCompositor.java

@@ -18,6 +18,7 @@ package androidx.media3.effect;
 import static androidx.media3.common.util.Assertions.checkNotNull;
 import static androidx.media3.common.util.Assertions.checkState;
 import static androidx.media3.common.util.Assertions.checkStateNotNull;
+import static androidx.media3.common.util.Util.contains;
 import static java.lang.Math.abs;
 import static java.lang.Math.max;
 
@@ -26,6 +27,7 @@ import android.opengl.EGLContext;
 import android.opengl.EGLDisplay;
 import android.opengl.EGLSurface;
 import android.opengl.GLES20;
+import android.util.SparseArray;
 import androidx.annotation.GuardedBy;
 import androidx.annotation.IntRange;
 import androidx.annotation.Nullable;
@@ -45,7 +47,6 @@ import com.google.common.collect.ImmutableList;
 import com.google.common.collect.Iterables;
 import java.io.IOException;
 import java.util.ArrayDeque;
-import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
 import java.util.Queue;
@@ -78,7 +79,8 @@ public final class DefaultVideoCompositor implements VideoCompositor {
 
   private static final String THREAD_NAME = "Effect:DefaultVideoCompositor:GlThread";
   private static final String TAG = "DefaultVideoCompositor";
-  private static final int PRIMARY_INPUT_ID = 0;
+  // TODO: b/338579287: Use the first registered index instead of a constant value.
+  private static final int PRIMARY_INPUT_INDEX = 0;
 
   private final VideoCompositor.Listener listener;
   private final GlTextureProducer.Listener textureOutputListener;
@@ -88,7 +90,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
   private final VideoFrameProcessingTaskExecutor videoFrameProcessingTaskExecutor;
 
   @GuardedBy("this")
-  private final List<InputSource> inputSources;
+  private final SparseArray<InputSource> inputSources;
 
   @GuardedBy("this")
   private boolean allInputsEnded; // Whether all inputSources have signaled end of input.
@@ -124,7 +126,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
     this.settings = settings;
     this.compositorGlProgram = new CompositorGlProgram(context);
 
-    inputSources = new ArrayList<>();
+    inputSources = new SparseArray<>();
     outputTexturePool =
         new TexturePool(/* useHighPrecisionColorComponents= */ false, textureOutputCapacity);
     outputTextureTimestamps = new LongArrayQueue(textureOutputCapacity);
@@ -142,25 +144,26 @@ public final class DefaultVideoCompositor implements VideoCompositor {
   }
 
   @Override
-  public synchronized int registerInputSource() {
+  public synchronized void registerInputSource(@IntRange(from = 0) int inputIndex) {
-    inputSources.add(new InputSource());
+    checkState(!contains(inputSources, inputIndex));
-    return inputSources.size() - 1;
+    inputSources.put(inputIndex, new InputSource());
   }
 
   @Override
-  public synchronized void signalEndOfInputSource(int inputId) {
+  public synchronized void signalEndOfInputSource(int inputIndex) {
-    inputSources.get(inputId).isInputEnded = true;
+    checkState(contains(inputSources, inputIndex));
+    inputSources.get(inputIndex).isInputEnded = true;
     boolean allInputsEnded = true;
     for (int i = 0; i < inputSources.size(); i++) {
-      if (!inputSources.get(i).isInputEnded) {
+      if (!inputSources.valueAt(i).isInputEnded) {
         allInputsEnded = false;
         break;
       }
     }
 
     this.allInputsEnded = allInputsEnded;
-    if (inputSources.get(PRIMARY_INPUT_ID).frameInfos.isEmpty()) {
+    if (inputSources.get(PRIMARY_INPUT_INDEX).frameInfos.isEmpty()) {
-      if (inputId == PRIMARY_INPUT_ID) {
+      if (inputIndex == PRIMARY_INPUT_INDEX) {
         releaseExcessFramesInAllSecondaryStreams();
       }
       if (allInputsEnded) {
@@ -168,7 +171,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
         return;
       }
     }
-    if (inputId != PRIMARY_INPUT_ID && inputSources.get(inputId).frameInfos.size() == 1) {
+    if (inputIndex != PRIMARY_INPUT_INDEX && inputSources.get(inputIndex).frameInfos.size() == 1) {
       // When a secondary stream ends input, composite if there was only one pending frame in the
       // stream.
       videoFrameProcessingTaskExecutor.submit(this::maybeComposite);
@@ -177,12 +180,13 @@ public final class DefaultVideoCompositor implements VideoCompositor {
 
   @Override
   public synchronized void queueInputTexture(
-      int inputId,
+      int inputIndex,
       GlTextureProducer textureProducer,
       GlTextureInfo inputTexture,
       ColorInfo colorInfo,
       long presentationTimeUs) {
-    InputSource inputSource = inputSources.get(inputId);
+    checkState(contains(inputSources, inputIndex));
+    InputSource inputSource = inputSources.get(inputIndex);
     checkState(!inputSource.isInputEnded);
     checkStateNotNull(!ColorInfo.isTransferHdr(colorInfo), "HDR input is not supported.");
     if (configuredColorInfo == null) {
@@ -196,10 +200,10 @@ public final class DefaultVideoCompositor implements VideoCompositor {
             textureProducer,
             inputTexture,
             presentationTimeUs,
-            settings.getOverlaySettings(inputId, presentationTimeUs));
+            settings.getOverlaySettings(inputIndex, presentationTimeUs));
     inputSource.frameInfos.add(inputFrameInfo);
 
-    if (inputId == PRIMARY_INPUT_ID) {
+    if (inputIndex == PRIMARY_INPUT_INDEX) {
       releaseExcessFramesInAllSecondaryStreams();
     } else {
       releaseExcessFramesInSecondaryStream(inputSource);
@@ -225,11 +229,11 @@ public final class DefaultVideoCompositor implements VideoCompositor {
   }
 
   private synchronized void releaseExcessFramesInAllSecondaryStreams() {
-    for (int i = 0; i < inputSources.size(); i++) {
+    for (int inputIndex = 0; inputIndex < inputSources.size(); inputIndex++) {
-      if (i == PRIMARY_INPUT_ID) {
+      if (inputIndex == PRIMARY_INPUT_INDEX) {
         continue;
       }
-      releaseExcessFramesInSecondaryStream(inputSources.get(i));
+      releaseExcessFramesInSecondaryStream(inputSources.valueAt(inputIndex));
     }
   }
 
@@ -241,7 +245,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
    * began.
    */
   private synchronized void releaseExcessFramesInSecondaryStream(InputSource secondaryInputSource) {
-    InputSource primaryInputSource = inputSources.get(PRIMARY_INPUT_ID);
+    InputSource primaryInputSource = inputSources.get(PRIMARY_INPUT_INDEX);
     // If the primary stream output is ended, all secondary frames can be released.
     if (primaryInputSource.frameInfos.isEmpty() && primaryInputSource.isInputEnded) {
       releaseFrames(
@@ -292,7 +296,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
       return;
     }
 
-    InputFrameInfo primaryInputFrame = framesToComposite.get(PRIMARY_INPUT_ID);
+    InputFrameInfo primaryInputFrame = framesToComposite.get(PRIMARY_INPUT_INDEX);
 
     ImmutableList.Builder<Size> inputSizes = new ImmutableList.Builder<>();
     for (int i = 0; i < framesToComposite.size(); i++) {
@@ -313,7 +317,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
     textureOutputListener.onTextureRendered(
         /* textureProducer= */ this, outputTexture, outputPresentationTimestampUs, syncObject);
 
-    InputSource primaryInputSource = inputSources.get(PRIMARY_INPUT_ID);
+    InputSource primaryInputSource = inputSources.get(PRIMARY_INPUT_INDEX);
     releaseFrames(primaryInputSource, /* numberOfFramesToRelease= */ 1);
     releaseExcessFramesInAllSecondaryStreams();
 
@@ -333,18 +337,18 @@ public final class DefaultVideoCompositor implements VideoCompositor {
     if (outputTexturePool.freeTextureCount() == 0) {
       return ImmutableList.of();
     }
-    for (int inputId = 0; inputId < inputSources.size(); inputId++) {
+    for (int i = 0; i < inputSources.size(); i++) {
-      if (inputSources.get(inputId).frameInfos.isEmpty()) {
+      if (inputSources.valueAt(i).frameInfos.isEmpty()) {
         return ImmutableList.of();
       }
     }
     ImmutableList.Builder<InputFrameInfo> framesToComposite = new ImmutableList.Builder<>();
     InputFrameInfo primaryFrameToComposite =
-        inputSources.get(PRIMARY_INPUT_ID).frameInfos.element();
+        inputSources.get(PRIMARY_INPUT_INDEX).frameInfos.element();
     framesToComposite.add(primaryFrameToComposite);
 
-    for (int inputId = 0; inputId < inputSources.size(); inputId++) {
+    for (int i = 0; i < inputSources.size(); i++) {
-      if (inputId == PRIMARY_INPUT_ID) {
+      if (i == PRIMARY_INPUT_INDEX) {
         continue;
       }
       // Select the secondary streams' frame that would be composited next. The frame selected is
@@ -353,7 +357,7 @@ public final class DefaultVideoCompositor implements VideoCompositor {
       //   2. Two or more frames, and at least one frame has timestamp greater than the target
       //      timestamp.
       // The smaller timestamp is taken if two timestamps have the same distance from the primary.
-      InputSource secondaryInputSource = inputSources.get(inputId);
+      InputSource secondaryInputSource = inputSources.valueAt(i);
       if (secondaryInputSource.frameInfos.size() == 1 && !secondaryInputSource.isInputEnded) {
         return ImmutableList.of();
       }

libraries/effect/src/main/java/androidx/media3/effect/MultipleInputVideoGraph.java

@@ -33,6 +33,7 @@ import android.opengl.EGLContext;
 import android.opengl.EGLDisplay;
 import android.opengl.EGLSurface;
 import android.util.SparseArray;
+import androidx.annotation.IntRange;
 import androidx.annotation.Nullable;
 import androidx.media3.common.C;
 import androidx.media3.common.ColorInfo;
@@ -75,7 +76,7 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
   private final Executor listenerExecutor;
   private final VideoCompositorSettings videoCompositorSettings;
   private final List<Effect> compositionEffects;
-  private final List<VideoFrameProcessor> preProcessors;
+  private final SparseArray<VideoFrameProcessor> preProcessors;
 
   private final ExecutorService sharedExecutorService;
 
@@ -114,7 +115,7 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
     this.compositionEffects = new ArrayList<>(compositionEffects);
     this.initialTimestampOffsetUs = initialTimestampOffsetUs;
     lastRenderedPresentationTimeUs = C.TIME_UNSET;
-    preProcessors = new ArrayList<>();
+    preProcessors = new SparseArray<>();
     sharedExecutorService = newSingleThreadScheduledExecutor(SHARED_EXECUTOR_NAME);
     glObjectsProvider = new SingleContextGlObjectsProvider();
     // TODO - b/289986435: Support injecting VideoFrameProcessor.Factory.
@@ -136,7 +137,7 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
   @Override
   public void initialize() throws VideoFrameProcessingException {
     checkState(
-        preProcessors.isEmpty()
+        preProcessors.size() == 0
             && videoCompositor == null
             && compositionVideoFrameProcessor == null
             && !released);
@@ -211,10 +212,10 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
   }
 
   @Override
-  public int registerInput() throws VideoFrameProcessingException {
+  public void registerInput(@IntRange(from = 0) int inputIndex)
-    checkStateNotNull(videoCompositor);
+      throws VideoFrameProcessingException {
-
+    checkState(!contains(preProcessors, inputIndex));
-    int videoCompositorInputId = videoCompositor.registerInputSource();
+    checkNotNull(videoCompositor).registerInputSource(inputIndex);
     // Creating a new VideoFrameProcessor for the input.
     VideoFrameProcessor preProcessor =
         videoFrameProcessorFactory
@@ -223,7 +224,7 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
                 // Texture output to compositor.
                 (textureProducer, texture, presentationTimeUs, syncObject) ->
                     queuePreProcessingOutputToCompositor(
-                        videoCompositorInputId, textureProducer, texture, presentationTimeUs),
+                        inputIndex, textureProducer, texture, presentationTimeUs),
                 PRE_COMPOSITOR_TEXTURE_OUTPUT_CAPACITY)
             .build()
             .create(
@@ -254,17 +255,16 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
 
                   @Override
                   public void onEnded() {
-                    onPreProcessingVideoFrameProcessorEnded(videoCompositorInputId);
+                    onPreProcessingVideoFrameProcessorEnded(inputIndex);
                   }
                 });
-    preProcessors.add(preProcessor);
+    preProcessors.put(inputIndex, preProcessor);
-    return videoCompositorInputId;
   }
 
   @Override
-  public VideoFrameProcessor getProcessor(int inputId) {
+  public VideoFrameProcessor getProcessor(int inputIndex) {
-    checkState(inputId < preProcessors.size());
+    checkState(contains(preProcessors, inputIndex));
-    return preProcessors.get(inputId);
+    return preProcessors.get(inputIndex);
   }
 
   @Override
@@ -285,7 +285,7 @@ public abstract class MultipleInputVideoGraph implements VideoGraph {
 
     // Needs to release the frame processors before their internal executor services are released.
     for (int i = 0; i < preProcessors.size(); i++) {
-      preProcessors.get(i).release();
+      preProcessors.get(preProcessors.keyAt(i)).release();
     }
     preProcessors.clear();
 

libraries/effect/src/main/java/androidx/media3/effect/PreviewingSingleInputVideoGraph.java

@@ -109,6 +109,6 @@ public final class PreviewingSingleInputVideoGraph extends SingleInputVideoGraph
 
   @Override
   public void renderOutputFrame(long renderTimeNs) {
-    getProcessor(SINGLE_INPUT_INDEX).renderOutputFrame(renderTimeNs);
+    getProcessor(getInputIndex()).renderOutputFrame(renderTimeNs);
   }
 }

libraries/effect/src/main/java/androidx/media3/effect/SingleInputVideoGraph.java

@@ -16,11 +16,13 @@
 
 package androidx.media3.effect;
 
+import static androidx.media3.common.util.Assertions.checkArgument;
 import static androidx.media3.common.util.Assertions.checkState;
 import static androidx.media3.common.util.Assertions.checkStateNotNull;
 
 import android.content.Context;
 import androidx.annotation.Nullable;
+import androidx.media3.common.C;
 import androidx.media3.common.ColorInfo;
 import androidx.media3.common.DebugViewProvider;
 import androidx.media3.common.Effect;
@@ -38,9 +40,6 @@ import java.util.concurrent.Executor;
 @UnstableApi
 public abstract class SingleInputVideoGraph implements VideoGraph {
 
-  /** The ID {@link #registerInput()} returns. */
-  public static final int SINGLE_INPUT_INDEX = 0;
-
   private final Context context;
   private final VideoFrameProcessor.Factory videoFrameProcessorFactory;
   private final ColorInfo outputColorInfo;
@@ -56,6 +55,7 @@ public abstract class SingleInputVideoGraph implements VideoGraph {
   private boolean isEnded;
   private boolean released;
   private volatile boolean hasProducedFrameWithTimestampZero;
+  private int inputIndex;
 
   /**
    * Creates an instance.
@@ -86,6 +86,7 @@ public abstract class SingleInputVideoGraph implements VideoGraph {
     this.renderFramesAutomatically = renderFramesAutomatically;
     this.presentation = presentation;
     this.initialTimestampOffsetUs = initialTimestampOffsetUs;
+    this.inputIndex = C.INDEX_UNSET;
   }
 
   /**
@@ -99,9 +100,11 @@ public abstract class SingleInputVideoGraph implements VideoGraph {
   }
 
   @Override
-  public int registerInput() throws VideoFrameProcessingException {
+  public void registerInput(int inputIndex) throws VideoFrameProcessingException {
     checkStateNotNull(videoFrameProcessor == null && !released);
+    checkState(this.inputIndex == C.INDEX_UNSET);
 
+    this.inputIndex = inputIndex;
     videoFrameProcessor =
         videoFrameProcessorFactory.create(
             context,
@@ -159,11 +162,11 @@ public abstract class SingleInputVideoGraph implements VideoGraph {
     if (outputSurfaceInfo != null) {
       videoFrameProcessor.setOutputSurfaceInfo(outputSurfaceInfo);
     }
-    return SINGLE_INPUT_INDEX;
   }
 
   @Override
-  public VideoFrameProcessor getProcessor(int inputId) {
+  public VideoFrameProcessor getProcessor(int inputIndex) {
+    checkArgument(this.inputIndex != C.INDEX_UNSET && this.inputIndex == inputIndex);
     return checkStateNotNull(videoFrameProcessor);
   }
 
@@ -193,6 +196,10 @@ public abstract class SingleInputVideoGraph implements VideoGraph {
     released = true;
   }
 
+  protected int getInputIndex() {
+    return inputIndex;
+  }
+
   protected long getInitialTimestampOffsetUs() {
     return initialTimestampOffsetUs;
   }

libraries/effect/src/main/java/androidx/media3/effect/VideoCompositor.java

@@ -15,6 +15,7 @@
  */
 package androidx.media3.effect;
 
+import androidx.annotation.IntRange;
 import androidx.media3.common.ColorInfo;
 import androidx.media3.common.GlTextureInfo;
 import androidx.media3.common.VideoFrameProcessingException;
@@ -46,29 +47,35 @@ public interface VideoCompositor extends GlTextureProducer {
   }
 
   /**
-   * Registers a new input source, and returns a unique {@code inputId} corresponding to this
+   * Registers a new input source.
-   * source, to be used in {@link #queueInputTexture}.
+   *
+   * @param inputIndex The index of the input source which could be used to determine the order of
+   *     the input sources. The same index should to be used in {@link #queueInputTexture}. The
+   *     index must start from 0. All inputs must be registered before {@linkplain
+   *     #queueInputTexture(int, GlTextureProducer, GlTextureInfo, ColorInfo, long) queueing}
+   *     textures.
    */
-  int registerInputSource();
+  void registerInputSource(@IntRange(from = 0) int inputIndex);
 
   /**
    * Signals that no more frames will come from the upstream {@link GlTextureProducer.Listener}.
    *
-   * @param inputId The identifier for an input source, returned from {@link #registerInputSource}.
+   * @param inputIndex The index of the input source.
    */
-  void signalEndOfInputSource(int inputId);
+  void signalEndOfInputSource(int inputIndex);
 
   /**
    * Queues an input texture to be composited.
    *
-   * @param inputId The identifier for an input source, returned from {@link #registerInputSource}.
+   * @param inputIndex The index of the input source, the same index used when {@linkplain
+   *     #registerInputSource(int) registering the input source}.
    * @param textureProducer The source from where the {@code inputTexture} is produced.
    * @param inputTexture The {@link GlTextureInfo} to composite.
    * @param colorInfo The {@link ColorInfo} of {@code inputTexture}.
    * @param presentationTimeUs The presentation time of {@code inputTexture}, in microseconds.
    */
   void queueInputTexture(
-      int inputId,
+      int inputIndex,
       GlTextureProducer textureProducer,
       GlTextureInfo inputTexture,
       ColorInfo colorInfo,

libraries/exoplayer/src/main/java/androidx/media3/exoplayer/video/CompositingVideoSinkProvider.java

@@ -419,7 +419,6 @@ public final class CompositingVideoSinkProvider
       outputColorInfo =
           inputColorInfo.buildUpon().setColorTransfer(C.COLOR_TRANSFER_ST2084).build();
     }
-    int videoGraphInputId;
     try {
       videoGraph =
           previewingVideoGraphFactory.create(
@@ -435,12 +434,12 @@ public final class CompositingVideoSinkProvider
         Size size = currentSurfaceAndSize.second;
         maybeSetOutputSurfaceInfo(surface, size.getWidth(), size.getHeight());
       }
-      videoGraphInputId = videoGraph.registerInput();
+      videoGraph.registerInput(/* inputIndex= */ 0);
     } catch (VideoFrameProcessingException e) {
       throw new VideoSink.VideoSinkException(e, sourceFormat);
     }
     state = STATE_INITIALIZED;
-    return videoGraph.getProcessor(videoGraphInputId);
+    return videoGraph.getProcessor(/* inputIndex= */ 0);
   }
 
   private boolean isInitialized() {

libraries/transformer/src/androidTest/java/androidx/media3/transformer/DefaultVideoCompositorPixelTest.java

@@ -744,7 +744,8 @@ public final class DefaultVideoCompositorPixelTest {
                     textureBitmapReader,
                     videoCompositor,
                     sharedExecutorService,
-                    glObjectsProvider)
+                    glObjectsProvider,
+                    /* inputIndex= */ i)
                 .setEffects(effectsToApply.build())
                 .build();
         inputVideoFrameProcessorTestRunners.add(vfpTestRunner);
@@ -855,8 +856,9 @@ public final class DefaultVideoCompositorPixelTest {
         TextureBitmapReader textureBitmapReader,
         VideoCompositor videoCompositor,
         @Nullable ExecutorService executorService,
-        GlObjectsProvider glObjectsProvider) {
+        GlObjectsProvider glObjectsProvider,
-      int inputId = videoCompositor.registerInputSource();
+        int inputIndex) {
+      videoCompositor.registerInputSource(inputIndex);
       DefaultVideoFrameProcessor.Factory.Builder defaultVideoFrameProcessorFactoryBuilder =
           new DefaultVideoFrameProcessor.Factory.Builder()
               .setGlObjectsProvider(glObjectsProvider)
@@ -870,7 +872,7 @@ public final class DefaultVideoCompositorPixelTest {
                     textureBitmapReader.readBitmapUnpremultipliedAlpha(
                         outputTexture, presentationTimeUs);
                     videoCompositor.queueInputTexture(
-                        inputId,
+                        inputIndex,
                         outputTextureProducer,
                         outputTexture,
                         ColorInfo.SRGB_BT709_FULL,
@@ -884,7 +886,7 @@ public final class DefaultVideoCompositorPixelTest {
           .setTestId(testId)
           .setVideoFrameProcessorFactory(defaultVideoFrameProcessorFactoryBuilder.build())
           .setBitmapReader(textureBitmapReader)
-          .setOnEndedListener(() -> videoCompositor.signalEndOfInputSource(inputId));
+          .setOnEndedListener(() -> videoCompositor.signalEndOfInputSource(inputIndex));
     }
   }
 

libraries/transformer/src/main/java/androidx/media3/transformer/AudioSampleExporter.java

@@ -97,7 +97,7 @@ import org.checkerframework.dataflow.qual.Pure;
   }
 
   @Override
-  public AudioGraphInput getInput(EditedMediaItem editedMediaItem, Format format)
+  public AudioGraphInput getInput(EditedMediaItem editedMediaItem, Format format, int inputIndex)
       throws ExportException {
     if (!returnedFirstInput) {
       // First input initialized in constructor because output AudioFormat is needed.

libraries/transformer/src/main/java/androidx/media3/transformer/EncodedSampleExporter.java

@@ -129,7 +129,7 @@ import java.util.concurrent.atomic.AtomicLong;
   }
 
   @Override
-  public GraphInput getInput(EditedMediaItem item, Format format) {
+  public GraphInput getInput(EditedMediaItem item, Format format, int inputIndex) {
     return this;
   }
 

libraries/transformer/src/main/java/androidx/media3/transformer/SampleExporter.java

@@ -64,10 +64,11 @@ import java.util.List;
    *
    * @param editedMediaItem The initial {@link EditedMediaItem} of the input.
    * @param format The initial {@link Format} of the input.
+   * @param inputIndex The index of the input.
    * @throws ExportException If an error occurs getting the input.
    */
-  public abstract GraphInput getInput(EditedMediaItem editedMediaItem, Format format)
+  public abstract GraphInput getInput(
-      throws ExportException;
+      EditedMediaItem editedMediaItem, Format format, int inputIndex) throws ExportException;
 
   /**
    * Processes the input data and returns whether it may be possible to process more data by calling

libraries/transformer/src/main/java/androidx/media3/transformer/TransformerInternal.java

@@ -637,7 +637,7 @@ import org.checkerframework.checker.nullness.qual.MonotonicNonNull;
         }
 
         GraphInput sampleExporterInput =
-            sampleExporter.getInput(firstEditedMediaItem, assetLoaderOutputFormat);
+            sampleExporter.getInput(firstEditedMediaItem, assetLoaderOutputFormat, sequenceIndex);
         OnMediaItemChangedListener onMediaItemChangedListener =
             (editedMediaItem, durationUs, decodedFormat, isLast) -> {
               onMediaItemChanged(trackType, durationUs, isLast);

libraries/transformer/src/main/java/androidx/media3/transformer/TransformerMultipleInputVideoGraph.java

@@ -79,9 +79,9 @@ import java.util.concurrent.Executor;
   }
 
   @Override
-  public GraphInput createInput() throws VideoFrameProcessingException {
+  public GraphInput createInput(int inputIndex) throws VideoFrameProcessingException {
-    int inputId = registerInput();
+    registerInput(inputIndex);
     return new VideoFrameProcessingWrapper(
-        getProcessor(inputId), /* presentation= */ null, getInitialTimestampOffsetUs());
+        getProcessor(inputIndex), /* presentation= */ null, getInitialTimestampOffsetUs());
   }
 }

libraries/transformer/src/main/java/androidx/media3/transformer/TransformerSingleInputVideoGraph.java

@@ -106,12 +106,12 @@ import org.checkerframework.checker.nullness.qual.MonotonicNonNull;
   }
 
   @Override
-  public GraphInput createInput() throws VideoFrameProcessingException {
+  public GraphInput createInput(int inputIndex) throws VideoFrameProcessingException {
     checkState(videoFrameProcessingWrapper == null);
-    int inputId = registerInput();
+    registerInput(inputIndex);
     videoFrameProcessingWrapper =
         new VideoFrameProcessingWrapper(
-            getProcessor(inputId), getPresentation(), getInitialTimestampOffsetUs());
+            getProcessor(inputIndex), getPresentation(), getInitialTimestampOffsetUs());
     return videoFrameProcessingWrapper;
   }
 }

libraries/transformer/src/main/java/androidx/media3/transformer/TransformerVideoGraph.java

@@ -69,6 +69,8 @@ import java.util.concurrent.Executor;
    * <p>This method must called exactly once for every input stream.
    *
    * <p>If the method throws any {@link Exception}, the caller must call {@link #release}.
+   *
+   * @param inputIndex The index of the input, which could be used to order the inputs.
    */
-  GraphInput createInput() throws VideoFrameProcessingException;
+  GraphInput createInput(int inputIndex) throws VideoFrameProcessingException;
 }

libraries/transformer/src/main/java/androidx/media3/transformer/VideoSampleExporter.java

@@ -33,6 +33,7 @@ import android.media.MediaCodec;
 import android.media.MediaCodecInfo;
 import android.util.Pair;
 import android.view.Surface;
+import androidx.annotation.IntRange;
 import androidx.annotation.Nullable;
 import androidx.annotation.VisibleForTesting;
 import androidx.media3.common.C;
@@ -153,10 +154,10 @@ import org.checkerframework.dataflow.qual.Pure;
   }
 
   @Override
-  public GraphInput getInput(EditedMediaItem editedMediaItem, Format format)
+  public GraphInput getInput(EditedMediaItem editedMediaItem, Format format, int inputIndex)
       throws ExportException {
     try {
-      return videoGraph.createInput();
+      return videoGraph.createInput(inputIndex);
     } catch (VideoFrameProcessingException e) {
       throw ExportException.createForVideoFrameProcessingException(e);
     }
@@ -540,18 +541,19 @@ import org.checkerframework.dataflow.qual.Pure;
     }
 
     @Override
-    public int registerInput() throws VideoFrameProcessingException {
+    public void registerInput(@IntRange(from = 0) int inputIndex)
-      return videoGraph.registerInput();
+        throws VideoFrameProcessingException {
+      videoGraph.registerInput(inputIndex);
     }
 
     @Override
-    public VideoFrameProcessor getProcessor(int inputId) {
+    public VideoFrameProcessor getProcessor(int inputIndex) {
-      return videoGraph.getProcessor(inputId);
+      return videoGraph.getProcessor(inputIndex);
     }
 
     @Override
-    public GraphInput createInput() throws VideoFrameProcessingException {
+    public GraphInput createInput(int inputIndex) throws VideoFrameProcessingException {
-      return videoGraph.createInput();
+      return videoGraph.createInput(inputIndex);
     }
 
     @Override