zhenzhen-minimax-h3-ow-fast-video-lowprice

Comfly_minimax_h3_ow_fast_video_lowprice ("zhenzhen-minimax-h3-ow-fast-video-lowprice") is the fast tier of MiniMax's H3 video generation, and the model list tells you what "open world" gets you in one node:

• minimax-h3-ow-t2v-fast - text to video
• minimax-h3-ow-i2v-fast - image to video (first frame)
• minimax-h3-ow-r2v-fast - reference-to-video (multiple reference images for character/scene consistency)
• minimax-h3-ow-fl2va-audio-drive-fast and minimax-h3-ow-ref2va-audio-drive-fast - the audio-driven variants that animate a talking head / lip-sync digital human from an AUDIO input

That last pair is the interesting one: you drop in a video or reference images plus an audio clip, and H3 makes the mouth move. It's the kind of thing that normally means a whole separate pipeline, and here it's a dropdown.

It ships in the T8mars/Comfyui-zhenzhen pack, the "Comfly" API-wrapper collection that calls the author's paid reseller service. As a "lowprice" node it uses the cheaper China-facing endpoint (api.seedance.nz) and reads its key from the shared zhenzhen-seedance2-low-price-api-settings node through the api_config input. Submit, poll, download - same machinery as the rest of the pack, with results landing as a real VIDEO.

What you set

• model - required; pick your flavor. If you're doing character work with several reference shots, r2v-fast and ref2va are the ones.
• prompt - the scene description.
• seconds - 5, 10, or 15.
• resolution - 480p or 720p. This is the "fast/lowprice" tier, so 720p is the ceiling.
• ratio - the usual set (1:1, 2:3, 3:2, 3:4, 4:3, 9:16, 16:9, 21:9).
• image1–image9 - reference images for i2v/r2v/ref2va. For the audio-drive models, audio takes an AUDIO input.

skip_error and seed (cache control, not sent to the API) round out the optional list.

Outputs

• video - the VIDEO output, ready for a save or preview node.
• video_url - the hosted file as a string.
• task_id - for the site's async task list.
• response - raw payload for debugging.

Installing it

Whole pack, no models:

cd ComfyUI/custom_nodes
git clone https://github.com/T8mars/Comfyui-zhenzhen
# restart ComfyUI

Or ComfyUI Manager → "Comfyui-zhenzhen". Add the zhenzhen-seedance2-low-price-api-settings node with your key and wire api_config in.

The honest part

The "fast" tier is where the pack's value actually sits - these endpoints are priced to be the cheap way to prototype video, and the audio-drive variants are genuinely hard to get elsewhere in one node. The tradeoffs are the pack's usual ones: 720p cap, third-party reseller in the middle (your clips and reference images leave your machine), and upstream hiccups that rerun cleanly. And while MiniMax has an open-weights H3 for local use, the licensing famously excludes the US and EU - which is exactly why a cloud node like this stays relevant. If you're inside the excluded regions and want H3 at all, this is one of the few doors.