CivArchive
    Anirealism (Photo-Anime hybrid style) - SDXL beta
    Preview 66959117Preview 66959114Preview 66959112Preview 66959125Preview 66959118Preview 66959111Preview 66959116Preview 66959105Preview 66959107Preview 66959119Preview 66959113Preview 66959108Preview 66959137Preview 66959106Preview 66959115Preview 66959109Preview 66959121Preview 66959120Preview 66959122Preview 66959123

    I decided to create a style that creates people with anime proportions and faces but with realistic shading just for the fun of it. It works pretty well for cartoonish and surreal images that have a photorealistic aspect. It also seems to work with animals and vehicles as well, though it requires a bit more prompting to get them to work. I will continue updating this style if people like it to refine the look.

    Description

    Initial version of the model trained on SDXL using civitai's training service. The base dataset is 20 images of people created using 3d modeling software and then rendered on photorealistic HDR backgrounds. The trigger word for this version is just "anireal", and it seems to produce pretty consistent results on a variety of models. I recommend using DPM SDE++ 2M Karas on most SDXL models with a cfg of around 7 and about 30 steps, but it seems to work on turbo/lightning models as well at a lower step and cfg count. It works with a strength range of 0.6 to 1.0, but the effect falls off the lower it is. I recommend around 0.9 for the most flexibility while keeping the style consistent.

    LORA
    SDXL 1.0

    Details

    Downloads
    37
    Platform
    CivitAI
    Platform Status
    Available
    Created
    3/31/2025
    Updated
    9/30/2025
    Deleted
    -
    Trigger Words:
    anireal

    Files

    AnirealismXL-000006.safetensors

    Mirrors

    Huggingface (1 mirrors)