CivArchive
    Iridescent Hair - LORA - WIP - V1
    NSFW
    Preview 98760
    Preview 98759
    Preview 98758
    Preview 98757
    Preview 98756
    Preview 98755
    Preview 98754
    Preview 98753
    Preview 98752
    Preview 98751

    I would like to offer a wide variety of options when it comes to creating characters (more customization), from clothes, to hair, to irises and even hairstyles.

    In this Lora training I focused on creating iridescent hair, although due to my lack of knowledge (I only trained once) the character's eyes, clothes and even the face are affected by this effect, not just the hair.

    If you can help me I would appreciate it, here I leave the info of what I used:

    <lora:iridescent_hair:1> works quite well between 0.4 and 0.8

    It is trained with 35 images (Train Batch Size: 2 and Epoch: 1 = 1700 steps) birming clipped and clipped using the Kohya_SS utilities. LORA configuration:

    1- https://prnt.sc/qObBnzLo0by5

    2- https://prnt.sc/5IL9Fd_i1QJm

    This post is to reveal the first of many models that I want to share with the community. In fact, I feel that I can get better potential from my 3060 and that is why I would like you to send me a better configuration for Kohya_SS. If someone can also advise me on how to improve my training using Kohya in the comments, I would appreciate it.

    Description

    FAQ

    Comments (12)

    Daedalus_7Feb 13, 2023· 2 reactions
    CivitAI

    This is exactly the kind of content I want to see! Also, how did you make the cool gif? Thanks!

    DitamAi
    Author
    Feb 13, 2023· 1 reaction

    I did it in After Effect. https://prnt.sc/bOli7H-r0syA

    Daedalus_7Feb 13, 2023· 1 reaction

    Thanks for the info :)

    ritcher1Feb 22, 2023· 1 reaction
    CivitAI

    Recommended models with this great LORA ?

    DitamAi
    Author
    Feb 22, 2023· 1 reaction

    Abyss and AOM... Try anyone, I saw good results with everyone I tried. Some have even used realistic models and they look good.

    ritcher1Feb 22, 2023

    Thank you !

    DelsiginaFeb 25, 2023· 2 reactions

    I have around 30-40 checkpoints, flipping between several depending on my needs. I would say it depends on what your looking for. The main focus of this is hair, but Im testing the limites on what it can and cannot be applied to. Protogen and protogen fantasy work well. Some of my realistic furry / slime models are netting good results.

    DitamAi
    Author
    Feb 25, 2023· 1 reaction

    @Delsigina When I set it to 1, it appeared as I originally wanted it to appear, but it burned or pixelated, so to speak. That effect where everything looks SUPER FOCUSED. Example: https://prnt.sc/io2rlUbwCit0

    What I have learned is that if I get this super focused I can lower the CFG from 7 to 5 and get this result: https://prnt.sc/0sL7y5PzSxHI and if I upscale it, I get this: https://prnt.sc/_UOTkr60_qes

    DelsiginaFeb 25, 2023· 1 reaction

    @DitamAi epic pictures, even the "burned one" but thats not what I was posting about here lol. I was responding to @ritcher1 about what they are looking for. If they are doing certain things, some checkpoints are more favorable than others.

    marmelade500Mar 5, 2023
    CivitAI

    Can you tell me what app you used to animate the image with the falling feathers? Looks so cool.

    Zelous0918Mar 22, 2023· 1 reaction
    CivitAI

    This Lora is amazing, a bit overly tuned and can take over generation, if fine tuned would be my favorite Lora <3

    DitamAi
    Author
    Mar 22, 2023

    It was one of my first Lora... right now I have more knowledge... but they already released more efficient ones like lycoris, loha, LoCon and I don't know how to work with them. Q.Q do you know?

    LORA
    SD 1.5

    Details

    Downloads
    3,800
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/13/2023
    Updated
    8/6/2026
    Deleted
    -
    Trigger Words:
    glow hair
    multicolored hair

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.