OpenCut Studio
engineOpen editor
Script-to-video in one prompt

Edit video by writing it.OpenCut Studio turns a short prompt into a finished cut.

Upload clips, let the AI analyze the visuals and audio, paste a prompt, and watch the timeline assemble itself — track by track, beat by beat. Refine in place, then export.

Open the editorHow it worksengine
  • Deterministic mock providers, end-to-end
  • FFmpeg-powered scene detection
  • Multi-track timeline with banners & captions
Untitled project · autosaved
1920×1080 · 30fps
Prompt
30-second teaser · 3 scenes · call to action
Hook · 0:00–0:04Demo · 0:04–0:18Testimonial · 0:18–0:25CTA · 0:25–0:30
video
HookDemoProofCTA
overlay
BannerLower-third
caption
4 tracks · 7 clips · 30s · 12.4 MB est.ready to render

Multi-track timeline, not a magic trick

Each script beat lands on a real clip with start, duration, source, and metadata. Edit anything by hand — the AI never hides the cuts.

Honest AI

Provider state is shown up-front. Mock providers are clearly labelled so you know what is deterministic scaffolding vs. a real LLM call.

Built for review

Render-ready scenes, banners, and captions come back as a typed bundle you can audit, override, or feed back into the prompt.

How it works

Five steps from raw footage to a finished cut.

Every step is auditable, and the underlying data is always editable — you are never locked into an AI black box.

  1. 01

    Upload clips

    Drop raw takes, screen recordings, and b-roll. OpenCut fingerprints every byte and tags them in the asset library.

  2. 02

    Analyze

    FFmpeg probes media, the vision and audio providers extract scenes, transcripts, and visual tags ready for matching.

  3. 03

    Paste script

    A short prompt becomes a structured script with beats, intent tags, mood cues, and estimated durations.

  4. 04

    Generate timeline

    Each beat is matched to the best scenes and laid out across video, overlay, and caption tracks. Banners and lower-thirds slot in.

  5. 05

    Edit & export

    Trim clips, swap alternates, adjust banners and captions, then export a render-ready bundle — no proprietary lock-in.

Capabilities

The editor first. The AI serves the cut.

Each surface area is independently useful, so you can lean on the AI where it helps and keep full manual control where it does not.

EditorTimelineAI engineExport
Editing
  • Per-clip trim, split, replace
  • Project library with autosave
  • Three-pane workspace
Timeline
  • Multi-track clips colored by scene
  • Banner & lower-third overlay tracks
  • Caption tracks when providers allow
AI engine
  • Deterministic script generation
  • Vision + transcription + embeddings
  • Live job progress and retries
Export
  • Render-ready timeline JSON
  • Auditable provider attribution
  • Banners and captions from prompts
Live now

Open the editor and assemble a 30-second cut before this page reloads.

The editor is the same one you would use in production. Mock providers ensure every interaction is reproducible; swap in real LLM, vision, and transcription credentials when you are ready.

Open editorengine
AI engine—
Providers
/health
…
/projects
GET · POST
/scripts
POST · LIST
/timeline
GET · BUILD