CivArchive
    Simple Lora Loader for Hunyuan Video - fast gguf
    Preview 49919307
    Preview 50096051
    Preview 49919309
    Preview 49919308

    A basic workflow for loading a lora to use with Hunyuan video. Modified from the comfyUI example here.

    Loras can be stacked as needed by adding more LoraLoaderModelOnly nodes.

    With version to use with GGUF quantized Hunyuan Video models - as requested

    Description

    Example with the fasthunyuan lora. Set Shift to 17 and guidance >6.

    FAQ

    Comments (9)

    0101ARTJan 6, 2025
    CivitAI

    I tried a similar method yesterday, it works well, I even managed to combine several loras during generation

    bellydreams
    Author
    Jan 6, 2025

    I'm pleasantly surprised by how well the loras can stack!

    TurboCoomerJan 6, 2025

    I stack loras same old way as in SD, just putting one after another and didnt encounter any problems still

    0101ARTJan 6, 2025

    @bellydreams True, I did not fully understand how to maintain a balance. Let's say between the character-lora and action-lora. If you figured it out, then please tell me. I tried increasing the strength of character lora to 1.3, and decreased strength of action lora to 0.6

    psspsspsspssspssJan 6, 2025· 4 reactions
    CivitAI

    Please include an HD screenshot of the workflow. I like to know what the workflow is before I download it. THX

    bellydreams
    Author
    Jan 7, 2025· 1 reaction

    It's exactly as described and nothing special tbh! nonetheless I've added screenshots.

    Ponder_StibbonsJan 14, 2025
    CivitAI

    I'm not sure what the hell is wrong with me, but I'm at a loss as to why my loras are not showing up in the loader node. I tried copying to several additional folders other than usual lora folder, to no avail. Am I missing a dedicated lora folder for hunyuan? There are zero answers to this anywhere, which makes me feel like I'm missing something incredibly, embarrassingly obvious.

    bellydreams
    Author
    Jan 17, 2025· 1 reaction

    I know that feeling lol. No need for a dedicated folder they should show up if they are in /models/loras. I can only bring up my own experience -

    If you have different paths to /comfyUI/models/loras/ then check your folder configuration especially if using a loader like Stablility Matrix /Pinokio. My old setup I'd have to replace a symlink to the model folder each time I updated comfyUI.

    And make sure you refresh comfyUI after loading a browser window (by clicking refresh/refresh node definitions/pressing r)

    Ponder_StibbonsJan 18, 2025

    Thank you very much for the advice. I had the feeling that this was the case, given the lack of info. I did give in recently and installed pinokio after I nuked everything by moving to a larger SSD and didn't feel like redoing 8 manual installs. After sorting out the path and python issues, strange problems like this persist. Sometimes problems just resolve after reloading for the nth time. Frustrating not knowing what I did to fix it. Most times I just leave it for a few days and come back to find it working... Anyway, thanks again.

    Workflows
    Hunyuan Video

    Details

    Downloads
    153
    Platform
    CivitAI
    Platform Status
    Available
    Created
    1/6/2025
    Updated
    5/14/2026
    Deleted
    -

    Files

    simpleLoraLoaderFor_fastGguf.zip

    Mirrors