Pick the hook frame of an ad with video inspect stills

Sume video inspect returns 8 mid-bin stills by default, or 1-24 at timestamps you choose, plus optional transcription. Find the best opening frame fast.

5 min readSume
All posts

Call video inspect on a Sume-hosted ad and it returns 8 mid-bin stills by default, or stills at 1-24 timestamps you pass in frames.at[]. Look at the first seconds side by side, choose the opening frame, then cut or regenerate. Optional transcribe: true adds a transcript. See Video inspect.

Frame options

Pass exactly one of at[] or fps. Stills default to jpeg; png is available.

Inspect frame programs, from the Sume docs (read 2026-10-03).
`frames` valueResult
omitted8 mid-bin stills (1 fps if the clip is under 8 s)
falseProbe only, no stills
{ at: [seconds] }1-24 explicit timestamps
{ fps: n }Sampling rate, capped at 24 stills

Cost

Probe and stills are billed by their Modal compute. Transcription is $0.01 per audio minute (list prices from Sume's catalog rate card (GET /v1/catalog), confirmed in the repo's catalog file read 2026-10-03). The default mode is sync, waiting up to 30 seconds.

Typical use

Sample frames at 0, 0.5, 1, 1.5 and 2 seconds of each hook variant, and compare which one shows the product or face fastest. Then use video trim to cut the winner.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume