The AI storyboard generator that listensSay the scene.See the storyboard.
Talk through your story out loud. Frameteller turns it into storyboard panels where every character stays on model, lets you fix any panel by voice, and exports an animatic timed to your own narration.
-
Shot 1Wide, 0:00
Night. A narrow alley, rain coming down hard.
-
Shot 2Medium, 0:05
Mira kneels. Under a crate, a little cat is glowing.
-
Shot 3Close-up, 0:10
Close on its eyes. They light up like lanterns.
Same yellow coat, same girl, every shot. Durations come from how long you spoke each line.
The live demo opens soon
We go live with a working product on Mon Nov 2, 12:00 PM ET. Anyone can try it.
Live demo: Mon Nov 2, 12:00 PM ET
Get the launch post on XVoice to storyboard to animatic, in four steps
Four of Meta's models each do one job in the pipeline.
Speak
Narrate the scene. Muse Voice Transcribe writes it down live and tells co-directors apart.
Direct
Muse Spark breaks your words into shots, framing and dialogue, and keeps one description per character.
Draw
Muse Image draws each panel from character references, so faces and outfits stay consistent.
Refine and export
Click anything in a panel and SAM 3.1 selects it to move or change. Export a timed animatic.
Made by talking
Three panels from The Lighthouse, a story we told out loud: no sketching, just narration. Ada and Pip the fox stay on model from the storm to the sunrise, and the animatic is timed to the narrator's voice.
Shot 1WideA tiny lighthouse on a rock, waves crashing all around it.
Shot 4MediumAda opens the window and Pip tumbles inside, shaking off the rain.
Shot 6WideAda and Pip watch the sunrise over the calm sea.
Build log
Live updatesWe're building this in public, Oct 26 to Nov 4. New entries appear here without reloading.
- Shipped
A home at frameteller.com
Frameteller now lives at frameteller.com, and the old address and www both lead there.
- Shipped
Talk for a minute, get an animatic
You can now talk through a scene for a minute and download an animatic of the storyboard, timed to your words and voiced by your own narration.
- Shipped
The studio is open
You can now open the studio in your browser, tell a story out loud, tap Next shot at each new moment, approve the cast and watch the storyboard draw itself.
- Shipped
It's on the internet
The backend now runs on Azure, so anyone can narrate a scene to it over the internet and get shots, a cast and first panels back.
- Shipped
Talk to it, get shots back
You can now narrate a scene to the backend and watch shots, character sheets and first panels come back while you are still talking.
- Shipped
Connected to Meta's models
The backend now talks to all four Meta models it needs, listening, directing, drawing and selecting, and each one passed a live test against the real service.
- Building now
Building the engine
Started the backend: story rules, cast approval and budget guards, all tested before any UI.
- Shipped
Proved the hard parts
Our test scenes kept the same girl and cat on model across every panel, one click selected her cleanly, and voice-style edits changed her coat without disturbing the rest of the picture.
- Shipped
Mapped the Meta APIs
Read the docs for live speech, image editing and segmentation, and confirmed the pipeline works end to end on paper.
- Shipped
Marketing page is live
This page is now deployed on Azure Static Web Apps from our public repo. Every push to main updates this build log within a minute.
- Shipped
Project kicked off
Picked the idea, wrote the build plan and put this page up. The public live demo goes up on Nov 2.
- Up next
Proving the hard parts first
Testing speech-to-shot-list, and whether one character stays recognisable across six panels.