Starting capture: wiring into the Publisher lifecycle

1. Using the custom capturer in the Publisher

Wherever you create your publisher (e.g., in MainActivity or a dedicated call manager), you pass an instance of the custom capturer into the Publisher.Builder:

After you build a Publisher and provide your MirrorVideoCapturer instance - all outgoing video for that publisher comes from your custom capturer instead of the default one.

2. Once you create a Publisher with your custom capturer and call session.publish(publisher), the SDK will call into your capturer’s startCapture() method.

The high‑level flow is as follows:

Check camera state: ensure the underlying camera resource is initialized and ready. Start capture loop: Open the camera (if not already open). Start a repeating capture request (Camera2) or start the preview pipeline (CameraX / Camera1). Handle “not ready yet” cases: If camera setup is still in progress, queue the actual start until the camera is opened (avoids race conditions). Error handling: If something is fundamentally wrong (e.g., startCapture called too early), throw or signal an error.

After startCapture() succeeds, your capturer starts receiving camera frames and pushing them to the SDK.