CivArchive
    KYOCEL MAX - Illustrious Anime Style - KYOCEL MAX_V1
    NSFW
    Preview 91731023
    Preview 91731082
    Preview 91731176
    Preview 91731215
    Preview 91731702
    Preview 91731756
    Preview 91731780
    Preview 91731796
    Preview 91731804
    Preview 91731914
    Preview 91731959
    Preview 94150412
    Preview 94879829
    Preview 94879825
    Preview 94879828
    Preview 94879826
    Preview 94879827

    V2.0 - A slightly cleaner node setup with more refined notes, additional prompt templates to go along with my wildcards - https://civarchive.com/models/1885843/kyocel-character-wildcards - and a few new LoRAs.

    The contrast controller LoRA by reakaakasky:
    https://civarchive.com/models/1523055/contrast-controller-ilnai

    Can be useful for any generation if you find even after the detailer and color your images are still a smidge flat, or maybe you want to go for that higher contrast look!

    The Stabilizer LoRA by reakaakasky:
    https://civarchive.com/models/971952/stabilizer-ilnai
    I feel this LoRA has a lot more potential than how I use it here, but when generating existing characters with danbooru tags, it can help reintroduce a little bit of that cinematic lighting and detail.

    V1.0 - After several months of work and a lot of trial and error - I have finally achieved the style I was looking for. I wanted hyper-detailed cel shading, little to no gradient shading, vibrant color, and decent-looking eyes and hands. I think I have succeeded. This workflow uses xquizitSDXLIllustrious_V3 as its backbone, along with some particular prompting, upscaling, and inpainting to achieve the final look.

    Aside from style, I think this is a fairly decent "complete" workflow for any model you may prefer. It includes text to image, wildcard text prompting, image to image upscale (tiled and non-tiled), as well as inpainting and metadata extraction.

    SFW images are more often closer to the "Kyocel" style, while NSFW images might take a few extra iterations to get rid of any gradient shading.

    I have left as many notes inside the workflow as I could think of to help one achieve similar results. In the grand scheme of things, I am a total noob, so any feedback or suggestions are much appreciated.

    Links to all models and LoRAs can be found in the notes inside the workflow.

    I hope to create a LoRA out of this eventually that will work with multiple models, but am severely limited by my MacOS hardware, minimal BUZZ, and inexperience with cloud-based GPU clusters.

    One can dream.

    I will continue to upload new versions as I refine this workflow to make it even more convenient and easy to use.

    Any questions - comment or shoot me a DM any time!

    Description

    FAQ

    Workflows
    Illustrious

    Details

    Downloads
    151
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/1/2025
    Updated
    9/14/2026
    Deleted
    -

    Files

    kyocelMAXIllustrious_kyocelMAXV1.zip

    Mirrors