Here is my workflow for HunyuanVideo - using the hunyuanvideo-wrapper.
The example prompt in the workflow is nsfw.
Upscaler model download:
https://github.com/Phhofm/models/releases/download/1xDeJPG_realplksr_otf/1xDeJPG_realplksr_otf.safetensors
ComfyUI_windows_portable\ComfyUI\models\upscale_models
HunyuanVideoWrapper download links:
Transformer and VAE (single files, no autodownload):
https://huggingface.co/Kijai/HunyuanVideo_comfy/tree/main
Go to the usual ComfyUI folders (diffusion_models and vae)
LLM text encoder (has autodownload):
https://huggingface.co/Kijai/llava-llama-3-8b-text-encoder-tokenizer
Files go to ComfyUI/models/LLM/llava-llama-3-8b-text-encoder-tokenizer
Clip text encoder (has autodownload)
Either use any Clip_L model supported by ComfyUI by disabling the clip_model in the text encoder loader and plugging in ClipLoader to the text encoder node, or allow the autodownloader to fetch the original clip model from:
https://huggingface.co/openai/clip-vit-large-patch14, (only need the .safetensor from the weights, and all the config files) to:
ComfyUI/models/clip/clip-vit-large-patch14
Description
HunyuanVideo-Wrapper
FAQ
Comments (14)
DownloadAndLoadHyVideoTextEncoder
Failed to import transformers.models.timm_wrapper.configuration_timm_wrapper because of the following error (look up to see its traceback): cannot import name 'ImageNetInfo' from 'timm.data' (D:\ComfyUI_windows_portable\venv\Lib\site-packages\timm\data\__init__.py)
for errors you can go to https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues
but if you have git (https://github.com/git-guides/install-git) installed - you can go to your models\LLM folder in the terminal
and write (replace with your path):
cd F:\ComfyUI\10_ComfyUI_windows_portable\ComfyUI\models\LLM
git clone https://huggingface.co/Kijai/llava-llama-3-8b-text-encoder-tokenizer
make sure all files are full downloaded ~15gb
and make sure your comfyui is up to date
@tawsLoRa my comfyui is up to date and i “ve already had the LLM model in my models for a long time. except that the employee node automatically tries to download
I had this solved by copypasting an updated version of timm inside ComfyUI-tbox "hardcopy" of timm
See the issue for reference
https://github.com/kijai/ComfyUI-HunyuanVideoWrapper/issues/272
Which folder does the upscaler go into? ESRGAN?
ComfyUI_windows_portable\ComfyUI\models\upscale_models
Thanks for sharing. Newb here. Do you happen to know why increasing batch count in queue is not working in this workflow? or it is just happening to me?
maybe you need to put the Seed after generate to "randomize" and not "fixed"
Any chance you'd have a good i2v workflow? This one is really great, love the layout.
I'd love to see any other workflows of yours too if you have any more to share. This one is so clean and easy to tweak without breaking the whole thing lol.
i can upload the i2v workflow but the quality of hunyuanI2V is not so impressing.. currently im playing around with wan2.1 i2v and it creates higher resolution videos and better fluid movement but also takes longer
maybe later today i clean up my wan i2v workflow and upload it also
@tawsLoRa thanks! I'm trying to run the i2v you loaded and have been able to figure out all the discrepancies except this one:
HyVideoModelLoader
Can't import SageAttention: No module named 'sageattention'
Not sure whats happening here and it has been difficult to google lol
@Fenn you need to isntall sageattention (its little bit difficult but faster) you can switch to sdpa instead (the "attention_mode" from the model node)