OpenMontage

An open-source agentic video production system that turns any AI coding assistant into a full video production studio.

🎬

Introduction

OpenMontage is the first open-source, agentic video production system, created by Calesthio. It transforms your AI coding assistant (Claude Code, Cursor, Copilot, Windsurf, Codex) into a full video production studio. Describe what you want in plain language, and the agent handles research, scripting, asset generation, editing, and final composition.

Unlike simple slideshow tools, OpenMontage produces real video videos — the agent builds a corpus from free stock footage and open archives, retrieves actual motion clips, edits them into a timeline, and renders a finished piece. It supports image-based animation, text-to-video generation, real footage montages, and full cinematic productions. The project is AGPLv3 licensed and hosted at github.com/calesthio/OpenMontage.

💻

Installation

OpenMontage requires Python 3.10+, FFmpeg, Node.js 18+, and an AI coding assistant.

Quick Setup

git clone https://github.com/calesthio/OpenMontage.git
cd OpenMontage
make setup

Manual Setup

python3 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt
cd remotion-composer && npm install && cd ..
python -m pip install piper-tts
cp .env.example .env

Prerequisites: FFmpeg (brew install ffmpeg or sudo apt install ffmpeg), Node.js 18+, and an AI coding assistant.

🔄

How It Works

OpenMontage transforms natural language prompts into finished videos through a multi-step agentic pipeline:

  1. Research — The agent researches your topic with live web search, gathering context, facts, and references
  2. Scripting — Writes a narrative script with scene descriptions, visual direction, and voiceover
  3. Asset Generation — Generates AI images, retrieves stock footage, creates motion clips via video generation APIs
  4. Audio — Writes and narrates the script using TTS with voice direction, finds royalty-free background music
  5. Composition — Edits everything into a timeline with transitions, subtitles, and effects
  6. Review — Runs a multi-point self-review: ffprobe validation, frame sampling, audio level analysis, and subtitle checks
  7. Rendering — Produces the final video file with Remotion or FFmpeg

Every creative decision gets your approval through the Backlot storyboard system, and every provider selection is scored across 7 dimensions with an auditable decision log.

📋

Pipelines & Composition

OpenMontage supports multiple composition modes and content sources:

ModeDescription
Image AnimationFLUX-generated images with Ken Burns, crossfade, particle overlays
Video GenerationVeo, Kling, or Runway motion clips from text prompts
Stock MontageReal footage from free archives edited into a timeline
Atelier (Bespoke)Custom composition from scratch — every scene crafted independently
Reference RemixStart from a YouTube video, Short, Reel, or TikTok as reference
🎞️

Backlot Living Storyboard

The Backlot is OpenMontage's real-time production dashboard that shows you what the pipeline is actually doing. Stages light up, the script lands as a screenplay page, scene cards shimmer while assets generate, and every provider decision and dollar spent is displayed. The storyboard serves as an approval gate — asset generation pauses on a scene-by-scene contact sheet so you approve the visuals before the render.

python -m backlot open                  # Library — every project on disk
python -m backlot open      # One production's live board

When a run is complete, use ▶ REPLAY RUN to replay the entire production from its timestamps, scrubbable end to end.

🔌

Provider Configuration

OpenMontage works with a wide range of AI providers for different stages of the pipeline. LLM providers handle research and scripting, image providers generate visuals, video providers create motion clips, and TTS providers handle narration. Providers are configured in the .env file and the system scores each provider selection across 7 dimensions with an auditable decision log.

  • LLM — OpenAI, Anthropic, Google, or any OpenAI-compatible endpoint for agent reasoning
  • Image — FLUX, Stable Diffusion, DALL-E, or gpt-image-1 for asset generation
  • Video — Veo, Kling (via fal.ai), Runway ML for motion clip generation
  • TTS — OpenAI TTS, Google Chirp3-HD, Piper TTS (local), Edge-TTS for narration
  • Music — Auto-sourced royalty-free music from free archives
  • Stock Footage — Free stock video archives for real footage montages