CivArchive
    WAN 2.2 Horse Lora I2v/T2v - Low Noise iv2/tv2
    NSFW

    This is a WAN 2.2 Iv2 and T2v horse character lora for generating "realistic horses" or having them walk into a scene, (inspired by Pony ) since wan 2.2 had little to no training on horses i decided to donate to the community. It's very simple type in the code word "H0RSE" at the beginning of the prompt then describe the horse. with the training data. "muscular, brown stallion with a black mane and tail. The stallion's large, erect, pink penis and testicles are visible. " Then describe how to enter the scene, "enters the frame from the right and stands in the middle of the frame". or enters the scene from the left etc. If something looks wrong in generation just say "the brown stallion pink penis is between the hind legs" that should fix things. most of the prompt should be placed at the beginning, you can mix the wording around if you want but that's what is recommend. you can try to test around with the training data captions, i haven't tried generating without it, but you could maybe just prompt "H0RSE" then do a light description but I'm not sure. I haven't done much testing but this seems to work just fine with wan lighting, 2 step high and 2 step low, both high noise and low noise at 1 str, cfg 1. if you're stacking loras and have issues you can try to lower the str but so far i haven't had any issues. Again, this isn't indented for malicious purposes so please try to keep the comment section PG rated and keep things mostly "having issues" related. also, if you like the lora drop a comment, let me know, and ill be glad to take any requests on any future loras as well.

    Description

    This is a WAN 2.2 Iv2 and T2v horse character lora for generating "realistic horses" or having them walk into a scene, (inspired by Pony ) since wan 2.2 had little to no training on horses i decided to donate to the community. It's very simple type in the code word "H0RSE" at the beginning of the prompt then describe the horse. with the training data. "muscular, brown stallion with a black mane and tail. The stallion's large, erect, pink penis and testicles are visible. " Then describe how to enter the scene, "enters the frame from the right and stands in the middle of the frame". or enters the scene from the left etc. If something looks wrong in generation just say "the brown stallion pink penis is between the hind legs" that should fix things. most of the prompt should be placed at the beginning, you can mix the wording around if you want but that's what is recommend. you can try to test around with the training data captions, i haven't tried generating without it, but you could maybe just prompt "H0RSE" then do a light description but I'm not sure. I haven't done much testing but this seems to work just fine with wan lighting, 2 step high and 2 step low, both high noise and low noise at 1 str, cfg 1. if you're stacking loras and have issues you can try to lower the str but so far i haven't had any issues. Again, this isn't indented for malicious purposes so please try to keep the comment section PG rated and keep things mostly "having issues" related.

    FAQ

    LORA
    Wan Video 2.2 I2V-A14B

    Details

    Downloads
    108
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    4/20/2026
    Updated
    4/29/2026
    Deleted
    4/21/2026

    Files

    H0RSE low 100 epoc.safetensors

    Mirrors

    TensorFiles (1 mirrors)