H3 Max Recast file types: mp4, mov, webm, m4v or GIF?
fal lists mp4, mov, webm, m4v and gif as Recast inputs. What to check before sending a clip to Sume: a public HTTPS URL, a rounded-up duration, one probe.

What fal's page lists for the input video
Recast is MiniMax H3 Max's person swap: it recasts the people in a video using reference photos while keeping the source motion, camera, cuts and audio. fal's model page, read 2026-10-03, lists five accepted video file types and the two prices.
The same page does not state a maximum length or file size. Sume's own catalog does state limits, so the table puts the two side by side.
| Item | fal page | Sume docs |
|---|---|---|
| Video file types | mp4, mov, webm, m4v, gif | No format list; a public HTTPS video URL |
| Length | Not stated | 5 to 30 seconds, no single shot over 15 seconds |
| Resolutions | 768p and 1080p | 768p default, 1080p on request |
| List price per second | $0.30 at 768p, $0.45 at 1080p | Same list, billed with Sume's 1.25 margin |
| Reference photos | Included at no extra charge | 1 to 4, one per new person |
What Sume asks of the clip
Sume takes the source as a URL, not an upload. The Media inputs page says video URLs must be fetchable public HTTPS addresses, and that localhost, private-network, signed or private URLs and mismatched content types are rejected before the job is accepted.
Sume also needs a number from you. A Recast request must carry duration, the source length in seconds rounded up, and the value must land between 5 and 30. The request is refused without it, and because the output keeps the source length, the number is also what Sume uses to price the job.
- mp4 is the common choice; mov, webm and m4v are on fal's list, so a clip in one of them is in scope on the vendor side.
- A GIF has no soundtrack, so there is no audio for Recast to keep. Use a real video file when sound matters.
- A clip behind a login or an expiring signed link will fail the public-URL check even when the file type is right.
Probe it once, then submit
Sume's Video inspect reads one clip that already lives on media.sume.com and returns probe facts: duration, size, rotation, fps, codecs and audio. A clip from elsewhere is brought in first with POST /v1/media-imports.
Round the probed duration up, check it is 5 to 30 seconds, and send that as duration with the same video_url you probed. If the clip is over 30 seconds, trim it first.
curl -X POST https://api.sume.com/v1/video-inspect \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: probe-recast-source-001" \
-d '{"video_url": "https://media.sume.com/artifacts/artf_demo/source.mp4"}'Where this leaves you
The vendor lists the file types, Sume lists the limits, and neither promises that every codec inside a container works. Treat the probe as the cheap test: if inspect reads the clip, you have a duration to send, and a real 768p run on a short clip is the proof for anything unusual.
Sources
Related posts
More in Models
- Hebrew and Georgian text to speech API: he and ka on Sume TTS
Cartesia lists Hebrew (he) and Georgian (ka) on Sonic 3.6, 3.5 and 3. What to send to Sume TTS, why the library has no voice for them, and what a script costs.
- Higgsfield Soul 2 custom_reference_id is not a Sume parameter
Higgsfield's Soul 2 API takes custom_reference_id for a saved character. Sume's Soul is text-only; for a consistent character use a model that takes references.
- Soul 2 lookbook batch of 4: Higgsfield Soul on Sume
Generate a four-image lookbook set per request with Higgsfield Soul 2: batch_size 1 or 4, 720p or 1080p, seven ratios, text-to-image only.
- Higgsfield Soul on Sume: text to image, 1 or 4 images, 720p or 1080p
Soul is a text-to-image row in Sume's image catalog: seven aspect ratios, no references, 1 or 4 images per call, 720p or 1080p. Limits and price.
Written by Sume