# Video production

Build a shot-based workflow with clear acceptance criteria, selective revisions and a delivery budget.

## Define a usable shot

Turn the creative brief into a shot list before calling a video model. For each shot, record subject, action, camera movement, duration, aspect ratio and continuity requirements. Define what makes the result usable in the edit: a completed action, recognizable subject, sufficient handles and the required framing.

## Produce and review by shot

1. Separate the script into shots with one clear action. Identify which shots need reference images or prepared start and end frames.
2. Select a model by its documented duration, reference and audio capabilities in the [video model directory](https://seedrouter.ai/video-generation).
3. Generate and save the shot with its task identifier, brief and references.
4. Review the whole clip, including the first and last frames. Mark the exact defect and time range when rejecting a result.
5. Revise only the affected shot. Assemble approved clips in your editing application and check continuity across cuts.

## Keep a delivery record

A shot record should contain its input version, generation settings, output asset and acceptance decision. Store the result before exposing a preview so opening a review screen does not submit a new generation. Retain alternate takes until the edit is approved.

The [short-film workflow](https://seedrouter.ai/blog/claude-opus-5-5-seedance-2-5-short-film) demonstrates how planning and video generation fit together. Use the [task API guide](https://seedrouter.ai/docs/api/tasks) for completed-result handling; an accepted submission is not a finished video.

## Estimate the cost of delivery

Start with [current prices](https://seedrouter.ai/pricing), then account for every generated take used in the job, editing and human review. Cost per generated second and cost per accepted shot answer different questions. Avoid offering a fixed production quote from the price of one successful request alone.

## What the API covers

SeedRouter supplies the selected text, image and video model calls. Shot storage, editing, sound mixing and final delivery belong to your application or production tools. Native audio in a particular video model is not a separate music or speech API.

## Related resources

- https://seedrouter.ai/models/seedance-2-5
- https://seedrouter.ai/models/kling-3-0
- https://seedrouter.ai/integrations/n8n
- https://seedrouter.ai/integrations/comfyui
- https://seedrouter.ai/industries/marketing
- https://seedrouter.ai/industries/film-animation
- https://seedrouter.ai/blog/claude-opus-5-5-seedance-2-5-short-film
- https://seedrouter.ai/blog/seedance-api