Edit: 8/16/26: Fixed bug where Audio 1 wasn't plugged in to H3 REF2VA node.
Profil d'exécution
Description de la source
Edit: 8/16/26: Fixed bug where Audio 1 wasn't plugged in to H3 REF2VA node.
Adds Comfy-Kitchen Attention Speed Up
Adds Temporal Upsampling De-Rope (fixes fast motion blur/smudge) [ https://github.com/matlowai/ComfyUI-MAINodes ]
Replaces RIFE with built in comfy node Film VFI
Adds Model Shift node
This is a pretty large update. The Temporal Upsampling 2nd Pass option improves action scenes drastically.
Commentaire généré par l’IA
Explication générée par l’IA à partir des détails de la source et de la configuration. Les suggestions sont clairement signalées.
Use the MiniMax H3 v2.0 workflow to create or edit videos from text, images, or reference material. It switches among T2V, I2V, FFLF, and REF2V paths, accepts optional image, audio, and video references or custom dialogue and music, and supports lip-synced video output.
For motion work, the publisher lists FFLF video extension, editing, and motion copying from reference videos, plus an optional temporal upsampling pass.
The v2.0 notes say the workflow replaces RIFE with FILM VFI for frame interpolation and adds a temporal upsampling de-rope option.
Use text for T2V or images for I2V; reference controls support up to four pictures, two audio inputs, and one video. The workflow outputs video.
Custom dialogue or music can be supplied for lip-sync in T2V, I2V, and REF2VA modes.
The page points to workflow notes for model and node links, and describes ComfyUI Manager as an installation option for nodes.
The publisher's setup notes call for ComfyUI CUDA 13.0 and say to reinstall Sage Attention after updating CUDA.
The listed node packs are comfyui-easy-use, comfyui-frame-interpolation, comfyui-kjnodes, comfyui-videohelpersuite, and rgthree-comfy.
Listed model files include minimax_h3_audio_vae_fp32.safetensors, minimax_h3_video_vae_fp16.safetensors, qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors, and taeh3.safetensors.
Other listed model files are film_net_fp32.pt, minimax_h3_fl2va_pruned_int8_convrot.safetensors, and minimax_h3_ref2va_pruned_int8_convrot.safetensors.
Choose the T2V/I2V/FFLF or REF2VA path with the model switch; the description also lists custom audio for lip-sync with dialogue or music.
Optional controls include a second temporal upsampling pass, image cropping and a maximum-megapixel setting in the image loader, and switchable reference slots for four pictures, two audio inputs, and one video.
The publisher says this workflow has been replaced by MINIMAX SEED HUNTER WORKFLOW; check that version before choosing this one.
Suggestion
The listing gives no download source for film_net_fp32.pt; confirm where to obtain it before setup.
Suggestion
Faut-il modifier cela ?
Connectez-vous pour envoyer une demande de modification.
Sources
1 sourceExtraits de sources
1 extraitSource context: 23177 downloads · Type Workflows · Base model MiniMax H3
Hardware and download-size estimates are not available yet.
Exigences
film_net_fp32.pt
Non résoluCheckpoint · Unknown
minimax_h3_audio_vae_fp32.safetensors
Fichier non vérifiéVAE · Hugging Face · Comfy-Org/MiniMax-H3 · MiniMax H3
Checkpoint · SAFETENSORS · Unknown
minimax_h3_ref2va_pruned_int8_convrot.safetensors
Fichier non vérifiéCheckpoint · SAFETENSORS · Unknown
minimax_h3_video_vae_fp16.safetensors
Fichier non vérifiéVAE · Hugging Face · Comfy-Org/MiniMax-H3 · MiniMax H3
qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
Fichier non vérifiéText encoder · Hugging Face · Comfy-Org/MiniMax-H3 · MiniMax H3
Checkpoint · Hugging Face · Kijai/MiniMax-H3-TAE · MiniMax H3 TAE
comfyui-easy-use
TrouvéPack de nœud · Registry
comfyui-frame-interpolation
TrouvéPack de nœud · Registry
Pack de nœud · Registry
comfyui-videohelpersuite
TrouvéPack de nœud · Registry
rgthree-comfy
TrouvéPack de nœud · Registry