CivArchive
    Sketchy style LoRA [Anima] [illustrous] - AnimaV1
    NSFW
    Preview 137550893
    Preview 137550870
    Preview 137550874
    Preview 137550890
    Preview 137550882
    Preview 137550863
    Preview 137550869
    Preview 137550873
    Preview 137851133
    Preview 137851158
    Preview 137851202
    Preview 137852809
    Preview 137550895
    Preview 137852807
    Preview 137550885
    Preview 137853216

    Pretty cool sketchy art style, and yes I misspelled the trigger word in my dataset... (only for Illustrious)

    Trigger word for Illustrious: Skt3tchy

    Trigger word for Anima: @Sk3tchy

    For Anima, use a LoRA strength between 0.4 and 1.0, depending on the look you're aiming for. Compared to Illustrious, Anima appears to capture the details of this dataset much more faithfully, despite both models being trained on the exact same dataset. It's also significantly more versatile, adjusting the strength has a major impact on the final result, giving you much finer control over the output.

    Overall, Anima reproduces the intended art style much more accurately, whereas Illustrious tends to produce a smoother, more softened interpretation, which you can still get in Anima by lowering the strength and with prompting (sketch, flat_color, etc).

    You can add the normal danbooru "sketch" trigger word as well if you want an even sketchier look, just make sure that you don't also have it in your negative prompt if you decide to use it!

    Illustrious version is trained on WAI-NSFW V16 so use that for the best results!

    Anima version is trained on Anima V1 base so use that for the best results!

    Description

    Trained on mix of 1024 and 1536 res. Anima seems to have learned the small details much better than Illustrious.

    FAQ

    Details

    Downloads
    189
    Platform
    CivitAI
    Platform Status
    Available
    Created
    7/23/2026
    Updated
    8/2/2026
    Deleted
    -
    Trigger Words:
    @Sk3tchy

    Files

    anima_sketchy.safetensors

    Mirrors

    CivitAI (1 mirrors)