Scripting
  • English
  • App Store

      • title: Screen Capture: Video, Audio and Timing

      tag: PRO

      Screen Capture: Video, Audio and Timing

      Attach addVideoOutput({maxFPS?, buffering?}) and addAudioOutput({source, buffering?}) before start when initial samples matter. Audio source is system or microphone. System audio requires configuration.capturesAudio; microphone requires the user's picker selection. The actual microphone format is independent of configured system sample rate/channel count.

      Each output exposes samples, a single-consumer async iterable, closed, getStatistics(), onError, and dispose(). Breaking iteration detaches that output, without stopping the session or disposing a sample retained by the script. Always dispose each delivered sample in finally. Copy methods return independent Data; disposing the original does not invalidate those copies.

      Video exposes hasImage, nullable format, frame metadata, toImage(), copyPixelPlanes() and dispose(). Status-only video samples may lack pixels: inspect hasImage first. Pixel planes preserve each native plane's width, height and bytesPerRow, including padding. Pixel format/FourCC, bit depth and attachments describe the actual buffer; do not assume RGBA, a fixed format or SDR.

      Audio exposes actual ASBD fields, optional channel-layout bytes, sampleCount, and copyAudioBuffers() returning native buffer data/channel counts. Do not assume interleaving, Float32 or a fixed microphone format. discontinuity: true marks known bridge delivery loss; null means native continuity is unknown. Check timestamps for source discontinuities and resynchronize downstream when needed.

      Default buffering: video capacity 1 / dropOldest, audio capacity 32 / error. Set explicit capacity and overflow: dropOldest | dropNewest | error as needed. Queues are bounded before dispatch to JS; native capture never waits for JS. maxFPS limits video delivery, not the native capture rate. Slow consumers drop or fail explicitly; native recordings do not use these JS queues. Scripts that retain samples control their own retained memory.

      Timing uses native CMTime fields: value/epoch are bigint, plus timescale, flags and nullable seconds. Preserve rational values when synchronizing. session.timing.currentTime() reports the stream clock when available; hostTimeSeconds is provided only when convertible. Do not infer A/V synchronization from callback arrival order. Observe onFrameStatusChanged, onActivityChanged, selection, sample metadata and output/session statistics for diagnostics.

      index.tsx

      Run it inside the main App; the system picker requires user interaction. The Stop / Cancel button and closing the script release its session.

      Open the documentation’s example entry to run the demo, or view the complete index.tsx.