CivArchive
    Preview 6853479

    This is a ❤️ merge of many awesome 🫶 models with intent to create a model that is good GREAT for AI 🎨🖌️Art, Illustration (Anime💦🫠) and ... 📸Photography.

    SDXL-Lightning version works BEST with EXACTLY 8 sampling steps at CFG scale 1 and with HEUNPP2 sampler + SGM_UNIFORM scheduler !!! (also EULER + do your own tests...)


    All authors of used resources (at the moment of this model creation) stated that we can use and SHARE their work (models) as merged models. And for that I'm very grateful to all of them! 😊💖

    The RedOlives v3.0 used resources:

    Artium: https://civarchive.com/models/216439/artium?modelVersionId=277071
    DreamShaper XL: https://civarchive.com/models/112902/dreamshaper-xl?modelVersionId=351306
    Juggernaut XL: https://civarchive.com/models/133005/juggernaut-xl?modelVersionId=348913
    AlbedoBase XL: https://civarchive.com/models/140737/albedobase-xl?modelVersionId=329420
    SDXL Unstable Diffusers: https://civarchive.com/models/84040/sdxl-unstable-diffusers-yamermix?modelVersionId=309729

    🛠️ The selection of merged models in RedOlives v3.0 and v2.0 is somewhat different, so test them both - because they are producing slightly different results!


    The RedOlives v2.0 used resources:

    BASE Models:
    DreamShaper XL: https://civarchive.com/models/112902?modelVersionId=302806 https://civarchive.com/models/112902?modelVersionId=251662 by Lykon: https://civarchive.com/user/Lykon

    ✨Juggernaut XL: https://civarchive.com/models/133005?modelVersionId=288982 by KandooAI: https://civarchive.com/user/KandooAI

    ✨Tamarin_XL: https://civarchive.com/models/235746/tamarinxl by antinoice: https://civarchive.com/user/antinoice

    ✨Imaginarium: https://civarchive.com/models/249973?modelVersionId=282087 by FrenzyX: https://civarchive.com/user/FrenzyX
    SDXXXL: https://civarchive.com/models/167764?modelVersionId=253250 by razzz: https://civarchive.com/user/razzz

    ✨Pyro's NSFW SDXL: https://civarchive.com/models/133808/pyros-nsfw-sdxl?modelVersionId=324958 by pyros_sd_models: https://civarchive.com/user/pyros_sd_models

    ✨AlbedoBase XL: https://civarchive.com/models/140737/albedobase-xl?modelVersionId=329420 by albedobond: https://civarchive.com/user/albedobond

    ✨Crystal Clear XL: https://civarchive.com/models/122822/crystal-clear-xl by WarAnakin: https://civarchive.com/user/WarAnakin

    ✨Realistic Stock Photo: https://civarchive.com/models/139565?modelVersionId=294470 by PromptSharingSamaritan: https://civarchive.com/user/PromptSharingSamaritan

    ✨LEOSAM's HelloWorld SDXL Base Model: https://civarchive.com/models/43977/leosams-helloworld-sdxl-base-model by LEOSAM: https://civarchive.com/user/LEOSAM


    and 2 LORAs:

    ✨Detail Tweaker XL: https://civarchive.com/models/122359?modelVersionId=135867 by w4r10ck: https://civarchive.com/user/w4r10ck

    ✨SDXL FaeTastic Details: https://civarchive.com/models/134338?modelVersionId=293991 by Faeia: https://civarchive.com/user/Faeia


    Generation: for simple images I'm using 16-24 steps and CFG 4.5 - 7.5, but for complex ones - 34-50 steps.

    Suggested Resources:

    The ComfyUI Workflow for image generation: https://civarchive.com/articles/3898/comfyui-workflow-with-a-facedetailer-simple-but-very-nice

    The Workflow that was used to create this merge: https://civarchive.com/articles/3897/merging-models-in-comfyui-easy-to-understand-workflow

    As you can see from the list all used models were selected exactly for the purpose stated in the beginning: Art, Illustration and Photorealism. I hope you will be able to produce some great imagery with it. Have Fun!

    Description

    new approach to selection of models to merge

    FAQ

    Comments (2)

    EmilyperversionsFeb 20, 2024
    CivitAI

    Also same errors as the rest seem to have. Whether using invoke, fooocus or mage.space. There is something seriously wrong here. Maybe the software you use is more forgiving. I would recommend also trying other software and you'll see it too.

    0l1v1aR0551
    Author
    Feb 21, 2024

    I'm running it in Fooocus, Automatic1111, Forge and mostly in ComfyUI - the answer is - TAESDXL - use it (Tiled SDXL as VAE), not the VAE from Huggingface (in ComfyUI - node: LoadVAE - select TAESDXL, everywhere else - no VAE or explicitly set TAESDXL in settings). Besides - this setting gives you ability to overcome VRAM issues, when you are trying to save big upscaled images, which is "must have" now.

    Checkpoint
    SDXL 1.0

    Details

    Downloads
    1,123
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/19/2024
    Updated
    4/30/2026
    Deleted
    -

    Files

    redolives_v30.safetensors

    Available On (1 platform)

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