Introducing Backlot: the living storyboard for OpenMontage

[calesthio]

Backlot is the new storyboard view for OpenMontage.

In simple terms: when an AI agent is making a video, Backlot shows the whole production as it happens. Instead of only reading chat messages or digging through files, you can watch the run move through stages, review the script, see scene cards fill in, check generated assets, spot approval gates, and play back the final render from one visual board.

The best part: you do not have to start it manually during normal video creation. When a production starts, the agent opens the Backlot board for you automatically.

Watch the production come alive

Stages light up, active work is visible, costs and decisions stay on the board, and generated assets appear scene by scene.

Review the video before it renders

Backlot turns the scene plan into a storyboard/filmstrip. You can inspect the visuals, narration, takes, timing, and missing assets before the final compose step.

Approval gates are obvious

When the pipeline needs a human decision, Backlot makes the pause visible. You can see what is waiting, review the artifact, and then respond in chat.

Every project gets a home

The library shows your productions across the machine: live, awaiting review, rendered, or still sparse.

Backlot is intentionally read-only for v1. The agent and tools still write the source of truth to disk; Backlot makes that truth visible. It is the production room for OpenMontage: what is happening, what is waiting, what was decided, what it cost, and what is ready to review.

This is a big step toward making AI video production feel less like a black box and more like a real creative workflow.

[ghost]

can you guide me to proceed

[calesthio]

Absolutely. Here is the shortest path to proceed.

1. Set up the repo locally:

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

If you are on Windows PowerShell or do not have make, use the manual setup from the README:

py -3 -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -r requirements.txt
cd remotion-composer
npm install
cd ..
python -m pip install piper-tts
Copy-Item .env.example .env

2. Try Backlot without generating a real video:

python scripts/backlot_simulate_run.py
python -m backlot open backlot-demo-run

That opens a simulated live production board so you can see stages, script, storyboard, gates, assets, and replay behavior.

3. For a real production, open this repo in your AI coding assistant and ask for a video, for example:

Make a 60-second animated explainer about how neural networks learn.

During a normal production run, the agent should initialize a project under projects/<project-id>/ and open Backlot automatically. If you want to open it yourself, use:

python -m backlot open                  # project library
python -m backlot open <project-id>     # one project board

4. Add API keys only for the providers you want to use. They go in .env. You can start with local/free paths, but keys like OPENAI_API_KEY, ELEVENLABS_API_KEY, FAL_KEY, PEXELS_API_KEY, or PIXABAY_API_KEY unlock more generation and stock-media options.

If you share what you want to make first, I can point you to the right pipeline and the minimum setup needed for that specific video.