ISO MP4 video loses elapsed timeline relative to its audio during live variable-rate Zoom recording. In the 180-second run at 05c35f8, ISO-01 reports last accepted/processed video PTS 180.155s, but ffprobe shows the final video packet ending at 176.971s while AAC ends at 180.181s. All eight ISO files show a 1.1–3.2 second difference; Program is within 5ms. This is actual packet timing, not only a container duration label.
Reproduction: eight Zoom ISO sources, source frames arriving at variable rates, file nominal 60fps, fragmented MF MP4 sink, six hardware and two software placements. Preserve capture timestamps and source media; do not hide the discrepancy with nominal FPS or repeated frames.
Investigating nominal sample durations at fragment boundaries. Local evidence: artifacts/live-2026-09-14/iso-capacity-summary.json and iso-capacity-b-finalized.json. Recorder fixes remain in draft PR #531; no affected candidate has been published.
ISO MP4 video loses elapsed timeline relative to its audio during live variable-rate Zoom recording. In the 180-second run at 05c35f8, ISO-01 reports last accepted/processed video PTS 180.155s, but ffprobe shows the final video packet ending at 176.971s while AAC ends at 180.181s. All eight ISO files show a 1.1–3.2 second difference; Program is within 5ms. This is actual packet timing, not only a container duration label.
Reproduction: eight Zoom ISO sources, source frames arriving at variable rates, file nominal 60fps, fragmented MF MP4 sink, six hardware and two software placements. Preserve capture timestamps and source media; do not hide the discrepancy with nominal FPS or repeated frames.
Investigating nominal sample durations at fragment boundaries. Local evidence: artifacts/live-2026-09-14/iso-capacity-summary.json and iso-capacity-b-finalized.json. Recorder fixes remain in draft PR #531; no affected candidate has been published.