VideoRecorder vs AVCaptureSession#

Scripting ships two camera APIs that overlap intentionally. Pick the one that matches the shape of your problem.

#TL;DR

#Concrete differences

VideoRecorder is a stateful singleton. It owns one capture session, one writer, one audio session, and a strict state machine (idle → preparing → ready → recording → ...). Most apps need exactly that. The trade-off: you cannot run two clips in parallel, attach extra outputs, or change the pipeline mid-flight.

AVCaptureSession is plain AVFoundation. Each instance is independent. You can construct multiple sessions, but only one will actually have access to the camera at any given moment — the OS arbitrates that.

If you start an AVCaptureSession while VideoRecorder is recording (or vice versa) you will get a runtime error from the second consumer. Build your UI to make these mutually exclusive — typically by deciding once at the top of the page which API to use.

#Migrating an existing VideoRecorder use case

Most VideoRecorder users do not need to migrate. Migrate if you find yourself:

• Calling VideoRecorder.session and reaching into AVFoundation by hand.
• Working around the state machine (e.g. wanting recording and paused to be flipped over multiple files).
• Needing AVCaptureMetadataOutput (QR scanning while recording).

Migration is a per-feature port: enumerate the device with AVCaptureDevice, wrap it with AVCaptureDeviceInput, attach AVCaptureMovieFileOutput, and replace state-listener calls with explicit promise chaining. See the Quick Start for the full pattern.

#Both at once

You can instantiate AVCaptureSession while VideoRecorder is idle. Just make sure to:

1. Call await VideoRecorder.reset() first if it was used earlier in the same script.
2. Call await session.stopRunning(); session.dispose() before letting the script return control to the user.