Skip to main content
Performance mode lets you film real actors performing a shot, then re-render it in any world you can imagine — the actors’ movement, timing, blocking and expressions are preserved, while the set, wardrobe and grade come from your project’s assets. It’s AI filmmaking with human acting kept fully in the loop.
Performance generation is available on every plan (including free accounts) and is paid for with credits, like all other generation.

​
How It Works

  1. Film a take — Record your actors performing the shot on any camera (a phone is fine). Keep it between 2 and 15 seconds, MP4 or MOV.
  2. Upload it — In Shot Studio’s First Frame tab, switch to Performance mode and upload the take. Automat extracts the first frame automatically.
  3. Restyle the first frame — Describe the world you want in the Restyle Description, using @ mentions to bring in your characters, props and wardrobe. The actors’ poses and framing are preserved; the environment, wardrobe and lighting are repainted.
  4. Generate the video — In the Video tab, switch to Performance. Your take drives the motion and your styled first frame drives the look. Duration is matched to your take automatically, and the take’s original audio — your actors’ real voices — can be kept with the Keep my take’s audio toggle. The video description is optional here: use it for extra scene context, not choreography.
How to access: Performance mode lives in two places — the Performance switch in the First Frame tab (upload + restyle) and the Performance switch in the Video tab (final video generation).

​
When to Use

  • You want real, directed human acting in an AI-generated world
  • Your shot needs precise timing, blocking or emotional nuance that text prompts can’t reliably produce
  • You want to preview a scene in costume and location you don’t have access to

​
Tips for Best Results

  1. Frame it like the final shot — The extracted first frame sets the composition, so film with the framing you want in the finished clip.
  2. Keep takes tight — Credits scale with duration; a focused 5–8 second take usually beats a long one.
  3. Light evenly — Clean, even lighting on the actors gives the restyle step the best chance of preserving likeness.
  4. Restyle in steps — If a big transformation drifts, generate the restyled frame a few times or simplify the changes.
  5. Skip the restyle when you don’t need it — With no restyled frame set as active, the video is generated from the take’s own first frame and your description.
  6. Keep the real voices — With “Keep my take’s audio” on, your actors’ actual dialogue delivery carries into the finished shot — often the most powerful part of the workflow.
  7. Use footage you have rights to — Upload only takes of yourself or actors who’ve consented.
Performance video uses Kling 3.0 Motion — a motion-control model built for transferring real performances onto styled frames. See the Performance-to-Video models guide for pricing and options.

Generate Video (Direct)

Previous: direct video generation

Dialogue and Lipsync

Next: add dialogue and lipsync