CoolFace
Datasetpublic

malcolmrey/workflows

ComfyUI Workflows Catalog Author: malcolmrey A curated collection of production-ready ComfyUI workflows (both programmatic API .json and visual UI graphs) for cutting-edge Diffusion Transformer models. MiniMax-H3 Workflows Full technical documentation and API client examples: MINIMAX_H3_API_WORKFLOWS_GUIDE.md Available Workflows in H3/: 1. Accelerated Standalone Generation: workflow_api_minimaxh3_fbc_refmod_turbo_8step.json —… See the full description on the dataset page: https://huggingface.co/datasets/malcolmrey/workflows.

sourceHugging Faceapache-2.0updated 9d agoView on Hugging Face
30likes2.7kdownloads
Dataset Card

ComfyUI Workflows Catalog

Author: [malcolmrey](https://huggingface.co/malcolmrey)

A curated collection of production-ready ComfyUI workflows (both programmatic API .json and visual UI graphs) for cutting-edge Diffusion Transformer models.


MiniMax-H3 Workflows

Full technical documentation and API client examples: [`MINIMAX_H3_API_WORKFLOWS_GUIDE.md`](./MINIMAX_H3_API_WORKFLOWS_GUIDE.md)

Available Workflows in H3/:

1. Accelerated Standalone Generation:
  1. 1.`workflow_api_minimaxh3_fbc_refmod_turbo_8step.json`Primary Standard (8-Step Euler, FBC Fast 0.10, LightX Turbo LoRA, RefMod persona conditioning). ~41s render per 5s video.
  2. 2.`workflow_api_minimaxh3_fbc_refmod_turbo_4step.json`High-Speed Prototyping (4-Step Euler, FBC Fast 0.10, LightX Turbo LoRA, RefMod persona conditioning). ~21s render per 5s video.
  3. 3.`workflow_api_minimaxh3_fbc_refmod_standard.json`Archival Base Quality (20-Step RES Multistep, FBC Safe 0.08, RefMod persona conditioning). ~110s render.
2. Accelerated C2V (Clip-to-Video) Continuous Chaining:
  1. 1.`workflow_api_minimaxh3_fbc_refmod_c2v_turbo_8step.json`Primary Movie Production C2V (8-Step Euler, FBC Fast 0.10, LightX Turbo LoRA, RefMods, 22-frame motion context carryover). ~41s per scene.
  2. 2.`workflow_api_minimaxh3_fbc_refmod_c2v_turbo_4step.json`Rapid C2V Storyboarding (4-Step Euler, FBC Fast 0.10, LightX Turbo LoRA, RefMods, 22-frame motion context carryover). ~21s per scene.
  3. 3.`workflow_api_minimaxh3_fbc_refmod_c2v_standard.json`Archival C2V Master Quality (20-Step RES Multistep, FBC Safe 0.08, RefMods, 22-frame motion context carryover).
3. Core & Legacy Workflows:
  1. 1.`workflow_api_minimaxh3_t2v.json` — Multimodal Text-to-Video with synchronized spoken dialogue and ambient acoustic generation.
  2. 2.`workflow_api_minimaxh3_i2v.json` — Image-to-Video animation.
  3. 3.`workflow_api_minimaxh3_r2v.json` — Multi-modal Reference-to-Video (combines reference images, audio, and video).
  4. 4.`workflow_api_minimaxh3_c2v.json` — Clip-to-Video legacy continuation.
  5. 5.`workflow_api_minimaxh3_ref_c2v.json` — Reference conditioning + Clip-to-Video legacy continuity.
  6. 6.`workflow_minimaxh3_refmod.json` — Interactive ComfyUI Web UI visual graph for RefMod persona testing.