MiniMax H3 Image-to-Video and First/Last-Frame ComfyUI Workflows (Turbo, 6 steps)
=================================================================================

Article: https://freakyjolly.com/best-minimax-h3-image-to-video-comfyui-workflow/

What's inside:
  - freakyjolly-minimax-h3-i2v-turbo6.json: ComfyUI workflow (UI format). Drag it onto the ComfyUI canvas or use Workflow > Open.
  - freakyjolly-minimax-h3-i2v-turbo6-api.json: Same graph in API format, for scripts and the /prompt endpoint.
  - freakyjolly-minimax-h3-fl2v-turbo6.json: ComfyUI workflow (UI format). Drag it onto the ComfyUI canvas or use Workflow > Open.
  - freakyjolly-minimax-h3-fl2v-turbo6-api.json: Same graph in API format, for scripts and the /prompt endpoint.

Required models (download each file and put it in the folder shown):
  - minimax_h3_fl2va_pruned-w4a8_convrot_pruned.safetensors
      folder: ComfyUI/models/diffusion_models/
      download: https://huggingface.co/Winnougan/MiniMax-H3-INT4_Convrot_ComfyUI/resolve/main/minimax_h3_fl2va_pruned-w4a8_convrot_pruned.safetensors
  - minimax_h3_turbo_4step_ema_ckpt500.safetensors
      folder: ComfyUI/models/loras/
      download: https://huggingface.co/larryvrh/MiniMax-H3-Turbo-Lora/resolve/main/minimax_h3_turbo_4step_ema_ckpt500.safetensors
  - qwen3vl_4b_int4_convrot.safetensors
      folder: ComfyUI/models/text_encoders/
      download: https://huggingface.co/Merserk/qwen3vl-4b-int4-convrot/resolve/main/qwen3vl_4b_int4_convrot.safetensors
  - mmh3-4b-ClipProj-v3.1-mlp.safetensors
      folder: ComfyUI/models/clip_projections/ (created by ComfyUI-ClipProj on first start)
      download: https://huggingface.co/NicoLab28/ClipProj-MiniMax-H3/resolve/main/mmh3-4b-ClipProj-v3.1-mlp.safetensors
  - minimax_h3_video_vae_int8_convrot.safetensors
      folder: ComfyUI/models/vae/
      download: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_int8_convrot.safetensors
  - minimax_h3_audio_vae_fp32.safetensors
      folder: ComfyUI/models/vae/
      download: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors

Custom nodes:
  - ComfyUI-MiniMax-H3-Turbo (MiniMaxH3TurboLoRA, MiniMaxH3TurboSampler): https://github.com/Larryvrh/ComfyUI-MiniMax-H3-Turbo  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)
  - ComfyUI-ClipProj (ClipProjLoader): https://github.com/nicolab28/ComfyUI-ClipProj  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)
  - ComfyUI-KJNodes (DiffusionModelLoaderKJ, PathchSageAttentionKJ, ModelPreviewOverrideKJ): https://github.com/kijai/ComfyUI-KJNodes  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)
  - ComfyUI-Easy-Use (easy cleanGpuUsed, easy clearCacheAll): https://github.com/yolain/ComfyUI-Easy-Use  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)
  - ComfyUI-VideoHelperSuite (VHS_VideoCombine): https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite  (ComfyUI Manager > Install Missing Custom Nodes, or git clone into ComfyUI/custom_nodes/)

How to use:
  1. Install the custom nodes above and restart ComfyUI.
  2. Download the six model files into the folders listed.
  3. Drag freakyjolly-minimax-h3-i2v-turbo6.json (one start image) or freakyjolly-minimax-h3-fl2v-turbo6.json (first + last frame) onto the canvas.
  4. Pick your image(s) in the "CHANGE ME" Load Image node(s) and set Resolution Selector to the aspect ratio closest to your image.
  5. Edit the prompt, set the duration and queue. Output: ComfyUI/output/video/MiniMax_H3_i2v* or MiniMax_H3_fl2v*.

Notes:
  - Sage Attention is set to "auto" in Patch Sage Attention KJ. If SageAttention is not installed (https://github.com/thu-ml/SageAttention), set it to "disabled".
  - ClipProjLoader device is cuda:0. Change it if your GPU has a different index.
  - Duration is set in seconds; the Math Expression node converts it to a frame count on H3's 17k+5 grid at 24 fps (5 s = 124 frames, 6 s = 158 frames).
  - MiniMax H3 has no negative prompt. Put exclusions ("No music, no captions") in the positive prompt.

Model weights are not included. Each model keeps its own licence; check the model page before commercial use.
Workflow shared by FreakyJolly.com.
