OpenMontage: Turn Your AI Coding Assistant Into a Video Production Studio
What if your AI coding assistant could produce entire videos — not just generate clips, but research topics, write scripts, source footage, edit timelines, and render finished productions?
OpenMontage makes this real. It’s the first open-source, agentic video production system, and it works with Claude Code, Cursor, Copilot, Windsurf, or any AI assistant that can read files and run code.
Not Just “Animate Some Images”
Most “AI video” tools animate still images and call it a day. OpenMontage can do that, but it can also:
- Build real-footage videos from free/open sources (Archive.org, Pexels, NASA, Wikimedia)
- Research your topic with 15-25+ web searches before writing a word
- Cut actual motion clips into edited timelines
- Render production-grade output with narration, music, captions, and transitions
The distinction matters. This isn’t slideshow-with-Ken-Burns. It’s actual video production.
12 Production Pipelines
| Pipeline | What It Produces |
|---|---|
| Animated Explainer | Research-backed explainers with AI visuals |
| Documentary Montage | Real footage from free/open archives |
| Cinematic | Trailers, teasers, mood pieces |
| Avatar Spokesperson | AI presenter videos |
| Talking Head | Speaker-led presentations |
| Clip Factory | Batch short-form clips from long content |
| Podcast Repurpose | Turn audio into video highlights |
| Screen Demo | Polished software walkthroughs |
| Localization & Dub | Translate and dub existing video |
| Animation | Motion graphics and kinetic typography |
| Hybrid | Mix real footage with AI-generated visuals |
Every pipeline follows the same structured flow:
research → proposal → script → scene_plan → assets → edit → composeReference-Driven Creation
Paste a YouTube video you like, and OpenMontage analyzes the transcript, pacing, scenes, keyframes, and style — then generates differentiated concepts based on your new topic.
“Here’s a YouTube Short I love. Make me something like this, but about quantum computing.”
You get back:
- What it keeps from the reference (pacing, hooks, structure)
- What it changes (topic, visuals, angle)
- Cost estimates before asset generation starts
- What it will actually look like with your available tools
No more “best guess prompt spaghetti.”
The Backlot: Live Production Dashboard
When a production runs, the Backlot opens automatically — a local board that fills itself as the pipeline executes. Stages light up, scripts appear as screenplay pages, scene cards shimmer during asset generation, and every provider decision and dollar spent is visible.
Storyboards become real approval gates. Asset generation pauses on scene-by-scene contact sheets so you approve visuals before the render, not after.
What It Actually Costs
Some example productions from the repo:
| Video | Duration | Cost |
|---|---|---|
| ”The Last Banana” (Pixar-style short) | 60 sec | $1.33 |
| ”VOID” product ad (OpenAI only) | 30 sec | $0.69 |
| ”Library at Alexandria” (history piece) | 70 sec | $0.02 |
| Ghibli-style anime animations | 30 sec | $0.15 |
The $1.33 Pixar short used Kling v3 for motion clips, Google Chirp3-HD narration, royalty-free music, word-level captions, and Remotion composition.
Zero-Cost Local Stack
No API keys? You still get:
- Piper TTS — Free offline narration
- Archive.org + NASA + Wikimedia — Free archival footage
- Pexels + Unsplash + Pixabay — Free stock (free API keys)
- Remotion — React-based video rendering
- HyperFrames — HTML/GSAP motion graphics
- FFmpeg — Encoding, subtitles, mixing
- Local GPU models — Wan2.1, Hunyuan, LTX, CogVideo
The free stack produces real videos, not demos.
Quick Start
git clone https://github.com/calesthio/OpenMontage.git
cd OpenMontage
make setupThen open in your AI coding assistant and prompt:
“Make a 60-second animated explainer about how neural networks learn”
Or for real-footage:
“Make a 75-second documentary montage about city life in the rain. Use real footage only, no narration, elegiac tone, with music.”
Why This Matters
Video production has been stuck in two modes: expensive professional tools or cheap AI toys that produce obviously-AI-looking output.
OpenMontage is the first system that treats video production as a workflow problem rather than a generation problem. The AI doesn’t just make clips — it orchestrates research, scripting, asset creation, editing, quality control, and delivery like a production team would.
700+ skill files teach the agent exactly how to use each tool. Every creative decision gets human approval. Every provider is scored across 7 dimensions before selection. Quality gates block slideshow-looking renders.
This is what agentic video production looks like.
Links: