Skip to content

Your first clip

This walkthrough takes you from a blank workspace to a finished rendered clip. It's the full production loop in miniature — everything else in this guide is a deeper look at one of these steps.

Before you start

Check that the tray icon is green and the status dots at the top of the studio are green too. If not, see Your studio machine.

1. Write and generate the voice

  1. Click the + button on the voice track of the timeline.
  2. Type a short line of dialogue — one sentence is perfect for a first try. (Write numbers as words: "zero", not "0".)
  3. Pick a voice and a tone preset from the dropdowns.
  4. Click Generate and listen.

Don't like the read? Click Generate again — every generation is a fresh take, and regenerating until you're happy is the normal way of working. When it sounds right, click Submit: the studio cleans up the audio and solves the lip sync, then places the clip on the timeline. See Voice.

2. Direct the performance

  1. Click anywhere on the timeline to jump to that moment — the avatar updates instantly.
  2. Open the gesture picker and hover over clips to preview them on the avatar. Add one — a shrug, a nod, the sunglasses move.
  3. Add a facial expression — a smile works beautifully.
  4. Press Play to watch it all blend together.

Scrub, adjust, and replay as much as you like — previewing costs nothing. See Composing a performance.

3. Set the camera and look

  1. Pick one of the three camera angles — Left, Center, or Right. For a camera move, frame the start and click Set First, then frame the end and click Set Last. See Camera.
  2. Pick the background for the clip — see Lighting & backdrop.

4. Record a take

  1. Pick an aspect ratio from the format dropdown (the default is 4:3). The frame overlay shows exactly what will be captured.
  2. Click Record. Since this is a fresh project, the studio asks you to name it first — call it "testing" or whatever suits.

The recording captures the whole performance into a take — this only takes a moment. The take appears in the Takes Browser; give it a recognisable name. See Recording and Projects & takes.

5. Render it

  1. In the Takes Browser, click the → button on your take to add it to the render queue.
  2. Click Run and confirm the switch to Render Mode.

The AI render is the one genuinely slow step — expect around 15–20 minutes for a ten-second clip, less for shorter ones. It runs two passes (the stylized look, then lip-sync refinement) and can be left working in the background.

6. Review and export

When it finishes, your clip appears in the Renders tab. Hover over the thumbnail to preview it. Happy? Select it and click Export to save the video to a folder.

Not quite right? Add the take to the queue again for a fresh attempt — renders vary between tries, just like the voice. See Rendering.

That's the whole loop: write → voice → direct → record → render → review. For final delivery, exported renders are upscaled to 4K — see Delivering in 4K.

Video placeholder

End-to-end walkthrough — first clip from blank to finished render (to be added).