One 30-second Wan take cut into three variants with Video trim

Generate one 30-second Wan 3.0 take at 480p, then cut three variants with Video trim at $0.02 each. Chain, cost and limits, read 2026-10-10.

4 min readSume
All posts

A 30-second Wan 3.0 take at 480p costs $1.8750, and three Video trim cuts add $0.06, so three usable variants cost $1.9350 in total. Wan accepts 2 to 30 seconds, which makes 30 the longest single take you can generate.

Cutting is cheaper than regenerating. Trim keeps [start, end) of one Sume-hosted clip and returns a new MP4, with no model call. If a take has three good moments, three trims turn them into three posts. Prices are from the Sume pricing code and the Video trim page (read 2026-10-10).

What is the endpoint chain?

One generation, then trims that can all run at once.

  • POST /v1/videos with wan-3.0, duration: 30, resolution: 480p, one prompt that describes a slow, continuous move.
  • POST /v1/video-trim three times: start: 2, duration: 8; start: 12, duration: 8; start: 22, duration: 8. Send only one of end or duration.
  • Each trim needs its own Idempotency-Key; the source file stays unchanged.
  • Optional: put each trim in a Timeline render with a voice-over, as in the voice-over pattern.

What does it cost?

The take is the main cost. Compare resolutions before choosing 480p.

30-second Wan 3.0 take plus three trims (read 2026-10-10)
ResolutionTake3 trimsTotal
480p$1.875$0.06$1.935
720p$3.750$0.06$3.810
1080p$7.500$0.06$7.560

What are the limits?

A long Wan take drifts: the subject on second 25 may not match second 3. Use prompts that avoid cuts and people entering the frame, and check the whole take before you trim. Trim has a precision option; read the Video trim page before relying on frame-exact cuts.

The source must be a workspace media.sume.com file, which a generated video already is. Trim returns a new artifact each time; the original stays. Trim does not mix audio or add a caption, so use Timeline 1.0 for that.

Wan 3.0 also accepts a first-frame image, so if the three variants must start from one product, lead with a still.

Is it better than three 8-second generations?

Three native 8-second jobs at 480p cost $1.50, against $1.935 for one 30-second take with three trims. Trimming only wins when you want the variants to look like the same scene, or when you do not know where the good moment is. For independent variants, regenerating is cheaper. Use A/B hooks from one body when the aim is hook testing.

Pre-publish checklist

Play the 30-second take twice before trimming. The best cut points are rarely at round numbers, and the three offsets in the example are only a starting guess.

Name each variant by its start offset in your own records so that you can tell which cut performed best.

  • Choose non-overlapping ranges.
  • Keep each variant under the limit of the platform you post to.
  • Download the originals; they are the durable copies.

Sources

Related posts

More in Use cases

All Use cases posts

Written by Sume