VideoEditor-For-Android Documentation

VideoEditor-For-Android is an Android-based video editor that provides core video processing capabilities. It is designed to work with Android's hardware-accelerated APIs and is not compatible with systems below Android 4.3.

Key features include:

• Video Capture: Uses Android APIs for video acquisition.
• Video Processing: Utilizes OpenGL for frame-level processing, including whitening, applying filters, and adding watermarks.
• Encoding/Decoding: Uses Android's MediaCodec for hardware-accelerated video encoding and audio/video separation.
• Audio Handling: Uses MediaCodec for audio mixing and processing.
• Editing Functions: Supports video recording, cutting, and adding background music (BGM).

The OutputSurface class manages a Surface that can be passed to MediaCodec.configure() to receive decoded video frames. It uses a SurfaceTexture to latch incoming frames and provides methods to render them using OpenGL ES.

Workflow

1. Initialize: Create an OutputSurface instance by passing a VideoInfo object. This prepares the GL environment and creates the underlying Surface.
2. Configure MediaCodec: Call getSurface() to retrieve the Surface and pass it to your MediaCodec configuration.
3. Process Frames:
   • Call awaitNewImage() to wait for and latch the next available frame from the SurfaceTexture.
   • Call drawImage() to render the latched texture onto the current EGL surface.
4. Cleanup: Call release() to discard EGL contexts and surface resources.

The InputSurface class manages the EGL state required to feed video frames into a MediaCodec encoder via an OpenGL ES 2.0 context.

To use it, you must first obtain a Surface from MediaCodec.createInputSurface(). The InputSurface then wraps this surface into an EGL window surface.

Workflow:

1. Initialize: Create an InputSurface instance by passing the Surface from your encoder.
2. Prepare Rendering: Call makeCurrent() to set the EGL context and surface for the current thread.
3. Render: Perform your OpenGL ES drawing commands.
4. Timestamp: Call setPresentationTime(long nsecs) to set the frame's timestamp in nanoseconds.
5. Publish: Call swapBuffers() to send the rendered frame to the video encoder.
6. Cleanup: Call release() to destroy the EGL context, surface, and the underlying Surface.

The InputSurface class provides the following public methods for managing the encoder input surface:

• InputSurface(Surface surface): Constructor. Takes a Surface obtained from MediaCodec.createInputSurface(). Throws NullPointerException if the surface is null.
• void makeCurrent(): Makes the EGL context and surface current on the calling thread. Throws RuntimeException if eglMakeCurrent fails.
• boolean swapBuffers(): Calls eglSwapBuffers to publish the current frame to the encoder.
• Surface getSurface(): Returns the underlying Surface that the MediaCodec receives buffers from.
• void setPresentationTime(long nsecs): Sends the presentation time stamp to EGL in nanoseconds using EGLExt.eglPresentationTimeANDROID.

The release() method must be called to prevent resource leaks. It performs the following actions:

1. If the current EGL context matches the InputSurface context, it clears the current context and surface using eglMakeCurrent with EGL_NO_SURFACE and EGL_NO_CONTEXT.
2. Destroys the EGL surface via eglDestroySurface.
3. Destroys the EGL context via eglDestroyContext.
4. Releases the underlying Surface via mSurface.release().
5. Nullifies all internal EGL and Surface references.

You can apply custom visual effects to the video stream by providing a GPUImageFilter implementation. When a filter is added, TextureRender will pass the frame through the filter's processing pipeline before final display.

Note: Adding a new filter will destroy the previous mGpuFilter if one exists.

The ICamera.Config class allows you to define hardware constraints and aspect ratios for the camera session.

Available fields:

• float rate: The aspect ratio (width/height). Default is 1.778f.
• int minPreviewWidth: The minimum required width for the camera preview.
• int minPictureWidth: The minimum required width for captured pictures.