MiniMax H3 OW 视频生成（3 合 1）

MiniMax H3 OW is the standard tier of MiniMax's H3 "Open-World" video line, and this node keeps it simple: three modes - T2V (text-to-video), I2V (image-to-video from one first frame), and R2V (reference-image-to-video) - behind one dropdown. There's no 9-image pool, no audio input, no video references. If you want a clean, predictable closed-model video render and don't need the multimodal circus, this is the node to grab. It's the quiet sibling to the pack's flashier Fast and H3 nodes.

Same pack deal: T8mars/ComfyUI_Seedance, API wrapper, key from api.seedance.nz, requests only.

How it works

Pick a model and the node shows exactly the inputs that mode uses. T2V takes just the prompt. I2V and R2V both need image1 connected (the tooltip is explicit: required for both, ignored by T2V) - the difference is that I2V animates that frame directly while R2V treats it as a reference for a fresh scene. Everything runs through the pack's standard upload → submit → poll → download pipeline, with the usual verified MP4 output.

The inputs that matter

• model - minimax-h3-ow-t2v, minimax-h3-ow-i2v, or minimax-h3-ow-r2v.
• prompt - required for T2V and R2V, optional for I2V (but a hint steers the animation).
• seconds - 5, 10 or 15.
• resolution - 480p or 720p.
• ratio - one of eight documented aspects (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9, 21:9).
• image1 - the required frame for I2V and R2V.

Outputs: video, video_url, task_id, response.

Installing it

ComfyUI Manager → search "ComfyUI Seedance", or:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/ComfyUI_Seedance.git
cd ../..
python -m pip install -r custom_nodes/ComfyUI_Seedance/requirements.txt

Restart, key from api.seedance.nz/console, Seedance API Config into api_config.

Common issues

• I2V/R2V won't run - image1 is required for both; T2V is the only mode that ignores it.
• T2V with an image connected - the image is ignored by T2V. No error, just no effect; switch modes or disconnect.
• 480p ceiling - this tier maxes at 720p. If you need 2K+ output, Hailuo H3 or FLUX 3 are the better picks in this pack.
• Prompt-quality drift - H3 OW responds well to the pack's Minimax_H3_Context_IR node: enhance your prompt first, then feed result_text into this node's prompt for a noticeably more directed clip.

It's the boring, dependable node - and that's a compliment. When you're iterating shot after shot and don't want surprises, this is the one that gets out of your way. For the faster, audio-driven, or reference-heavy variants, the pack's Minimax_H3_OW_Fast_Video node has those modes.