60-second short drama: Vidu Q4 $1.89 to $23.40, Sume from $7.64

The short answer

A 60-second episode built from ten 6-second shots costs 60 x $0.045 = $2.70 at Vidu Q4 Preview's 540P list rate, $1.89 with the 30% image-to-video discount, and $23.40 at 4K list ($16.38 discounted). Sume does not list Vidu. Built on Wan 3.0 at 720p, the same episode is 60 x $0.125 = $7.50 for the picture and $7.64 once dialogue and the final render are added. Vidu's release targets short-drama makers by name, so this is the format it is priced for.

The table

The last column adds the Sume extras that Vidu's per-second rate does not cover in this comparison: 900 characters of dialogue by TTS ($0.0475 per 1,000 characters = $0.04275) and one timeline render minute ($0.10, since renders bill per output minute, rounded up, with a one-minute minimum). The Vidu rows do not include any dialogue or assembly cost, since the release does not price them.

Takes are the real multiplier

Drama is dialogue-heavy and acting-dependent, so one take per shot is optimistic. At three takes per shot, the Wan 720p episode is 3 x $7.50 = $22.50 for the picture, Vidu 540P discounted is 3 x $1.89 = $5.67, and Vidu 4K discounted is 3 x $16.38 = $49.14. Only the keeper takes go on the timeline, but you pay for all of them.

Assembly on Sume

The Sume timeline takes up to 200 video slots and renders up to 30 minutes of output, so an episode with a cutaway every two seconds fits. Dialogue generated with TTS is placed as audio on the same timeline. The timeline docs list the request shape. If you generate the shots on Vidu and assemble on Sume, the picture cost is Vidu's and only the render and audio are Sume's.

Captions are a separate cost. Speech-to-text is $0.01 per audio minute, so captioning the 60-second episode adds $0.01, which does not change any total above at the precision shown. Watch for the minimum clip length instead: Seedance 2.0 accepts 4 to 15 seconds a clip, so a 3-second cutaway cannot be a standalone Seedance 2.0 job. Plan shot lengths against the row you pick, and trim on the timeline afterwards. A scene with ten 6-second shots avoids the problem on every row listed here.