The workflow

Every clip goes through the same stages, in order. Most of them are fast, cheap iteration — only the final render takes real time.

1. Write and generate voice — type the line in the Create Voice dialog, pick a voice and preset, hit Generate. Regenerate until you like the read.
2. Analyze — submit the audio so the studio can solve lip sync. Takes a few seconds; the timeline appears when it's done.
3. Compose — place gesture, expression, and head-orientation clips on the performance layers. Choose the look and set the camera.
4. Preview — scrub the timeline, play back, adjust. Repeat as much as needed — this costs nothing.
5. Record — when the performance is right, capture it into a take.
6. Queue and render — add the take to the render queue and run it through the AI render.
7. Review — check the finished render. Keep it, re-render, or discard.

For final delivery, keep going one step further: upscale the finished render to 4K.

The rhythm of working

The studio is built so that everything before the render is instant and repeatable. Don't aim for perfection in one pass — write, listen, scrub, adjust, and only record when it feels right. Takes are cheap; renders are the only thing worth being deliberate about.