CivArchive
    CAT - Citron Anime Treasure [Pony] - v3
    NSFW
    Preview 117412952
    Preview 117412942
    Preview 117412953
    Preview 117412948
    Preview 117412946
    Preview 117412951
    Preview 117412949
    Preview 117412944
    Preview 117412945
    Preview 117412943
    Preview 117412950
    Preview 117412947

    Follow me on X/Twitter for exclusive content: https://x.com/Citron_Legacy 😁

    Wield a rainbow's radiance,

    Create Colorful Anime Treasure with CAT!


    This Checkpoint (attempts to) bring Illustrious knowledge/power to Pony. The goal is to be compatible with Pony but also have some Illustrious magic.


    CAT PDXL v5 - (Pony Diffusion V6) x (Zankuro_Lora & Afrobull_Lora)

    CAT PDXL v4 - (CAT PDXL v2) x (Collection of my best Pony Loras)

    CAT PDXL v3 - (CAT IL v4:0.7) x (CAT PDXL v4:0.3)

    CAT PDXL v2 - (CAT_Illustrious_v10:0.2) x (WAI-ANI-PONYXL:0.8) - It also has the VAE from CAT_Illustrious_v10 baked into it.

    Description

    FAQ

    Comments (7)

    MetaGenJan 15, 2026· 3 reactions
    CivitAI

    you should train the background also for v5

    CitronLegacy
    Author
    Jan 17, 2026

    Ok. I have some ideas but I've never focused on getting good backgrounds so I'm not sure how it will go.
    v5 is going to focus more on stability/compatibility with Pony loras. Better backgrounds will have to wait for future versions.

    MetaGenJan 17, 2026· 1 reaction

    @CitronLegacy i heard a good technique is to mask characters out of some background training pictures and use both masked and unmasked in training data, but im still working on how to make good style loras so only regurgitating what i heard left and right x)

    CitronLegacy
    Author
    Jan 18, 2026· 1 reaction

    @MetaGen Oh nice that was the idea I had! This morning I started training a background lora (for Illustrious) where I had removed the characters so that the background was the focus and also had images with the characters.
    We'll see if the technique works or not!
    As for style Loras you just need enough images that emphasize the details you want. If you a style with flat colors you want lots of images of different things with flat colors. If you want a shiny style same thing but all images are shiny.
    My best style Loras are ones that I make the art and then remake it with image to image or use Lora output for a new version. My best styles I refine over several iterations but most styles are just one round trained on images that have what I want.

    MetaGenJan 18, 2026

    @CitronLegacy I'm trying to make a gravure style lora that focuses on poses and composition, but the training data makes it easy to bleed into asian girls with blurred background style lora XD

    NukichuuJan 16, 2026· 3 reactions
    CivitAI

    Thats one way to bring new life into old (dead? :C) loras!
    Pony with lora was so janky when it was hip and trendy :x

    CitronLegacy
    Author
    Jan 17, 2026

    Lol yeah its like beating a dead horse with another dead horse!
    It is very janky. Illustrious is 10x better at least. Now that I've started I keep thinking "oh maybe this would be the fix" so I keep trying ideas when the come up.

    Today I uploaded 40 Illustrious Loras and zero Pony Loras so thats a good sign that I prefer Illustrious lol

    Checkpoint
    Pony

    Details

    Downloads
    117
    Platform
    CivitAI
    Platform Status
    Available
    Created
    1/15/2026
    Updated
    6/17/2026
    Deleted
    -

    Files

    catCitronAnimeTreasure_v3.safetensors

    Mirrors