Turn any topic into a finished video
tofuvideo writes the script, generates the visuals, records the voiceover, times the captions, and edits the final cut. You pick the topic. Soft on effort. Sharp on results.
- Early access
- Free to try
- Runs on your own keys or fully local
New video
Rendering 2 of 3Scene 4 · Golden crust close-up
Only this scene re-renders — tweaks take seconds.
the whole pipeline, end to end
16:9 · 9:16 · 1:1 from one choice
runs fully local on your hardware
narrators with a point of view
Everything included
A full video team, in one tab
Scriptwriter, editor, voice artist, and motion designer — tofuvideo plays every seat, so a topic goes in and a publish-ready video comes out.
Scripts with a voice
AI writes the script and the scene beats. Pick from 10+ narrator personas — Doctor, Engineer, Storyteller, Energetic host — or design your own, no code required.
Scene-perfect visuals
Real stock photography for concrete subjects, AI imagery for abstract ideas — and the AI vision-picks the best on-topic photo for every scene.
Natural voiceover
Lifelike AI narration that doesn't sound like a GPS. Control the voice per scene when one moment needs a different energy.
Captions that pop
TikTok-style word-highlight captions, or classic CC with .srt/.vtt downloads. Fonts, colors, and highlight style live in the Caption studio.
Click-to-fix timeline
A Premiere-style timeline of scene blocks. Regenerate one scene's image or voice — only that scene re-renders, so tweaks take seconds.
Batch your calendar
Paste one topic per line or load a CSV, and queue a whole month of content in one sitting.
Music that ducks
Built-in beds auto-duck under the narration — the biggest “this isn't a raw AI video” tell. Or upload your own track.
Every format
16:9 for YouTube, 9:16 for Shorts, 1:1 for feeds — one choice drives composition end-to-end.
How it works
From topic to upload in four steps
You bring one sentence — the pipeline handles the script, scenes, voice, captions, and music from there.
- 1
Type a topic
Or paste your own script. Pick format, length, persona, and voice — that’s the whole brief.
- 2
AI drafts the story
Script plus a scene-by-scene breakdown, each scene tagged for real stock photo or AI art.
- 3
Scenes come alive
Visuals, voiceover, timed captions, and ducked music — all rendered in parallel.
- 4
Ship the cut
The final MP4 is assembled. Tweak any scene on the timeline and only that scene re-renders.
One flaky scene never kills the job — a 20-minute video always finishes.
Made for every feed
One topic. Three formats. Ten voices.
Pick YouTube, Shorts, or Square and the whole pipeline — framing, avatar, captions — adapts to fit.
YouTube 16:9
Shorts 9:16
Square 1:1
Narrator personas
The same script, told the way your audience listens.
Doctor — evidence-based and reassuring
Engineer — first-principles, loves a number
Professor — patient, builds ideas step by step
Energetic host — big hooks, zero dead air
Investigative journalist — digs in, follows the paper trail
Storyteller — opens with a scene, lands the point
- Create your own — describe the voice, no code
Straight from the engine
Real renders, unedited
No usage charts, no testimonials — this project is early. Instead, here are two actual MP4s the pipeline produced from a one-line topic. Script, visuals, voiceover, music, and captions: exactly as they came out.
“How to make cold brew coffee at home”
The full prompt. Everything else was the pipeline.
“Are black cats good as pets?”
The full prompt. Everything else was the pipeline.
Rendered with free-tier providers — local LLM script, stock + AI visuals, local voices. With paid keys plugged in, quality goes up from here.
FAQ
Questions, answered
Yes, fully. Every export is watermark-free and commercial use is included. Publish, monetize, or hand them to clients; they’re yours.
It depends on length and which providers you use, but scenes render in parallel so it’s minutes, not hours. Single-scene tweaks take seconds because only that scene re-renders — not the whole video.
You can upload your own background music today, and the per-scene rewrite lets you re-voice any line with a different AI voice or persona. Custom voice cloning is on the roadmap.
Click it on the timeline and regenerate just that image — or force AI generation instead of stock photography. The rest of the video is untouched, so a fix takes seconds, not a re-render.
Yes. Local mode runs the script, voice, and image steps on your own hardware with zero API spend — paid providers are skipped entirely. Or bring your own keys and pay providers directly, with no markup from us.
Export once per format: YouTube 16:9, Shorts/Reels/TikTok 9:16, or square 1:1. Captions, composition, and canvas adapt automatically — one choice drives the whole pipeline.
Yes — a full REST API: queue topics, poll render status, fetch the finished MP4. It’s the same API that powers the dashboard, so anything you can click, you can automate.
Your next video is one topic away
Type it. tofuvideo scripts, shoots, voices, captions, and cuts it. You take the credit.
Early access · Free to try · Your keys or fully local