Sync lip sync output: H.264 crf 17, HDR to SDR, alpha removed

Sync re-encodes all lip sync output to H.264 with libx264 at -crf 17 -preset slow regardless of the input codec, normalizes HDR to SDR, and removes alpha channels. Expect an SDR H.264 file whose bitrate and color grade may differ from your source. Sume's avatar video returns a final MP4 at a hosted media.sume.com URL.

Sync facts are from its Media Formats Support page; Sume facts from the Avatar videos, Runs and Audio detach docs, read 2026-10-01.

What does Sync's output look like?

How do I keep color closer to the source?

Sync's page suggests tagging SDR BT.709 metadata explicitly and, when color accuracy matters, exporting yuv444p, because its pipeline works in RGB. It adds that converting an existing 4:2:0 file to 4:4:4 cannot restore discarded chroma detail.

What does Sume hand back?

Avatar video burns optional captions into the clean final MP4 after generation, and resolution is currently 720p. Completed results carry public media.sume.com video artifacts, and the Runs docs say Format-run media URLs do not expire and are public to anyone holding the URL, so copy or proxy them if you need per-customer access. The docs I read do not state a codec or crf for the MP4, so inspect your own result if that matters. To pull the audio back out of a finished clip, Audio detach takes a Sume-hosted video URL and returns an audio_url.

What should I check before comparing outputs?

Compare like with like: resolution, frame rate and color tagging on both sides. For input rules on the Sume side see avatar video aspect ratios.