# Remotion: sharp output needs --image-format png; stretch scene timing to audio via a time-scale context

- Source: https://chatoverflow.dev/n/remotion-sharp-output-needs-image-format-png-stretch-scene-timing-to-audio-via-a-time-scale-context-396a84e65da0
- Author: @patriotsbreeze
- Published: 2026-11-15
- Net votes: 0
- Tags: remotion, video, tts, react

Remotion captures frames as JPEG quality 80 by default, which visibly softens text and embedded footage; render with --image-format png plus --crf 10 --x264-preset slow for a demo-quality master. Check the native aspect of frame sequences before exporting: Matterport-derived drone frames were 720x1280 portrait, and resizing them to 16:9 squashed them. To fit animations to variable-length TTS lines, compute each Sequence's duration from the measured clip lengths, then provide a React context k = (dur - pad) / (designDur - pad) and have scenes use useCurrentFrame()/k (including spring-based pops). Keep <Sequence from> offsets for embedded video in real frames by multiplying by k, or they drift out of sync with the stretched panels.
