OpenMontage Drama Studio: AI short-drama studio that renders stills on your own ComfyUI via the REST API

[charliewwdev]

Sharing OpenMontage Drama Studio, a self-hosted AI short-drama studio that can use your own ComfyUI server for image generation.

What it does: a one-line idea goes through 7 stages: script, characters, scenes, props, storyboard & shots, voice and edit. People use it for short dramas, viral short videos and e-commerce product ads.

How ComfyUI is used:

• Set COMFYUI_SERVER_URL. You can also set a per-capability URL such as COMFYUI_<CAPABILITY>_SERVER_URL. Character, scene, prop and storyboard images are then rendered on your ComfyUI instead of a paid API.
• The adapter (studio/server/providers/comfyui.py) uses only the standard REST API: POST /prompt, then polling GET /history, then fetching results with GET /view. It needs no custom nodes and has no SDK dependency.
• It reads object_info to find the installed checkpoints, diffusion models, text encoders and VAEs. On a stock SD1.5/SDXL install it builds a basic API-format checkpoint workflow. If a Qwen-Image stack is installed, it uses that instead.
• Prompts are mostly Chinese, and SDXL/RealVis CLIP barely reads CJK. So when the target model uses a Latin CLIP, the adapter rewrites the prompt to English first. If that rewrite isn't available, it falls back to a deterministic English-locked prompt.

Video shots use Volcengine Seedance and voice uses edge-tts. Seedream/Seedance are paid third-party APIs; ComfyUI is the self-hosted option for images. Deployment is Docker Compose with Postgres and Redis included.

• Repo: https://github.com/charliewwdev/OpenMontage
• Demo: https://openmontage.dtok.io

It's AGPL-3.0. It's an independent product rebuilt from the calesthio/OpenMontage codebase and isn't affiliated with that project.

[misakaai-dev]

Glad the image path uses ComfyUI.