Generation in invideo

Read /invideo first. Generate what the person's files and stock can't give (/invideo-media).

How to generate

Pick the model from the live list (invideo gen models / find_models); it changes as models ship.
Price it with invideo gen run … --dry-run / dry_run, say the cost, then run it with --max-credits / max_credits. Make one sample, review it, then make the rest.
With narration, generate the voice first, read its length (invideo media metadata / media), then generate picture to fit.
At most three tries per asset, changing something each time; then stop and tell the person. Redo only what failed.

The need	Read
Choosing and prompting a model	`references/models.md`
Model ids by job, the dry run and the retirement rule	https://docs.invideo.io/reference/models (markdown: add `.md`)
Voice-over, music and sound craft	`/invideo-media`
Checking what came back	`/invideo-media`