CivArchive
    Sydney Sweeney Hunyuan video Lora - v1.2

    Creates the likeness of the actress, Sydney Sweeney

    Guidance of 4-6 works best.

    v1.2 is trained at 512px with the keyword sw33n3y, it is trained with better data at a lower learning rate, and it significantly improves issues with glitching and frozen faces.

    v1.1 is trained at 1024px with a keyword (sw33ny) (yes I spelled it incorrectly, it was an accident)

    v1.0 is trained at 512.

    Description

    FAQ

    Comments (9)

    Juampab12Feb 6, 2025· 4 reactions
    CivitAI

    wow. the quality of the new version's videos are insane

    vfx_ai
    Author
    Feb 6, 2025· 1 reaction

    I KNOWWWWWWWWWWW, I'm going back and updating all of my old models

    zazaFeb 6, 2025

    @vfx_ai can you please share with us what is the training config that produced that high quality comapred to first and second lora

    vfx_ai
    Author
    Feb 6, 2025· 1 reaction

    @zaza 2e-5, captioned using llama3.2-vision:14b, the major difference is using at least 50 great images instead of 10-20

    zazaFeb 9, 2025

    @vfx_ai how many repeats, epoch or steps for 50 images

    Thank you.

    Conozco323Feb 6, 2025
    CivitAI

    such great work! do you consider requests or commissions? thanks!

    FluxFTWFeb 6, 2025
    CivitAI

    Wow, the sample videos are so good. Did you upscale them to add details or are they raw output?

    vfx_ai
    Author
    Feb 6, 2025

    raw output

    vfx_ai
    Author
    Feb 7, 2025

    I added some A/B testing videos with and without the LORA applied

    LORA
    Hunyuan Video

    Details

    Downloads
    774
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    2/6/2025
    Updated
    5/14/2026
    Deleted
    5/23/2025
    Trigger Words:
    sw33n3y

    Files

    sw33n3y-hunyuan-v1.2-vfx_ai.safetensors