I've been playing with MiniMax H3 in ComfyUI and kept bouncing between the canvas, a text editor for prompts, and a separate video editor. So I put together a custom node pack that keeps the whole loop in one sidebar, and I figured some of you might find it useful.What it does:<ul><li>Prompts you can actually inspect. You describe the shot and add references, exact dialogue or on-screen text. It builds a structured plan, a prompt and a validation report. If something is missing it tells you instead of guessing.</li><li>Official workflows, one click. It writes the official ComfyUI H3 workflow for your task mode (text, first frame, first + last frame, last frame, or a reference set) using the weights you already have installed. If you have your own H3 graph, it can connect the prompt into that instead, and leaves your models, LoRAs and sampler alone.</li><li>Clips collected as a project. Each run lands as a segment you can reorder, replace, regenerate and preview.</li><li>Longer videos. Write a storyboard script for up to 60 seconds. It splits it into 4–15 s segments, generates them in order and joins them.</li><li>A small built-in editor. It has up to eight tracks, trim/split, titles, picture and video overlays, crop, colour, cross dissolves and per-clip volume and fades. It renders a 1080p MP4.</li></ul>Optional: Ollama or Gemini can suggest prompt improvements, but nothing is applied unless you accept it. OpenAI and Anthropic connections currently support model discovery and readiness checks; prompt generation through those providers isn't enabled yet. The UI is in English, Traditional Chinese and Simplified Chinese.Installation and updates:The ZIP attached to this post is a snapshot. For the latest code, documentation and bug reports, use the <a target="_blank" rel="ugc" href="https://github.com/rookiestar28/ComfyUI-MiniMaxH3-Studio">ComfyUI-MiniMaxH3-Studio GitHub repository</a>.If your ComfyUI-Manager allows Git URL installation, open Manager → Install via Git URL and paste:<a target="_blank" rel="ugc" href="https://github.com/rookiestar28/ComfyUI-MiniMaxH3-Studio.git">https://github.com/rookiestar28/ComfyUI-MiniMaxH3-Studio.git</a>After installation, restart ComfyUI and refresh your browser. For future updates, use Manager's Update action for this pack, then restart and refresh again. If Git URL installation is unavailable in your Manager setup, the GitHub README includes manual Git installation instructions. <a target="_blank" rel="ugc" href="https://github.com/Comfy-Org/ComfyUI-Manager#how-to-use">ComfyUI-Manager documentation</a>Already installed the ZIP? Installing Manager won't turn that folder into a Git installation. To switch, save your workflows and download any videos you want to keep, stop ComfyUI, and move the old pack folder outside <code>custom_nodes</code>. Start ComfyUI again and install through Manager as described above. Keep only one copy of the pack inside <code>custom_nodes</code>.If you prefer to keep using ZIP files, choose Code → Download ZIP on GitHub and replace the entire existing pack folder with the new version. Avoid mixing files from different versions. Restart ComfyUI and refresh your browser afterward.A few honest caveats:<ul><li>The editor and rendering need Windows x64.</li><li>Longer-video generation is text-to-video only for now.</li><li>It's not a saved project format yet, so download what you render.</li><li>It's an independent project, not affiliated with MiniMax.</li></ul>Happy to hear feedback, bug reports, or how you'd use it differently. I'm especially curious whether the storyboard-to-segments flow makes sense to people who haven't seen it before.

Files with SHA256: ad8d35ac7698dcea6bf7d7dd0c84de3d94402fec03c8dd40e5cba714f9d50338