One 30 s micro-drama shot or three 10 s shots: same price, so split?

Seedance 2.5 pricing is linear in seconds, so one 30 second shot and three 10 second shots cost the same $17.33 at 720p and 9:16. Splitting does not save money on the first pass. It saves money on retries, because a bad moment costs you 10 seconds of regeneration instead of 30. Use one long shot only for continuous action.

This matters for micro-dramas, which Metricool reports as gaining traction, because the format is built from many short beats.

The price is linear

Seedance 2.5 accepts 4 to 30 seconds per clip. Its token formula is width x height x seconds x 24 / 1024, billed at the documented rate with the 1.25 margin. At 720p and 9:16 that works out to $0.5778 per second.

Why a split helps

A redo is a new paid job. In the three shot plan, redoing one shot adds $5.78. In the one shot plan, it adds the whole $17.33. If you expect to redo one shot in three, splitting is the cheaper plan.

A split also lets you vary the framing, since each clip gets its own prompt and first frame.

Why a split can hurt

Each seam is a place where the lead can change. Chain shots by extracting the last frame with video frames and passing it as the next clip's start image. Without that, a face or a coat may differ across the cut.

Each seam also costs time in the edit. Timeline 1.0 joins up to 200 slots, but you are the one choosing the cut points.

A rule of thumb

Pick the length by what is happening on screen, not by the price.

• Continuous move or one long take: one 15 to 30 second shot.
• Dialogue exchange with reactions: 6 to 10 second shots.
• A reveal or a cliffhanger: its own 4 to 6 second shot, so you can redo it alone.
• Unsure about the prompt: render the first 4 seconds as a test, since 4 seconds is the minimum.

What Sume does not do

Sume does not promise that two clips generated from the same prompt will match. It does not extend a clip for you; chaining is done by passing frames. Retries are billed as new jobs, as with every generation here.