CivArchive
    Preview 37933560
    Preview 37933977

    VAUXZ ANIME MERGE

    This model was merged to create a model that could generate:

    1. crisp details

    2. high contrast

    3. good dark vibes

    4. great lighting and depth

    5. Male characters as good as it does female characters

    6. great 832 x 1216 anime images

    Model was merged with my loras

    • Sampler: Euler a - Karras

    • Resolution: 832 x 1216, 1024 x 1024

    • CFG: 7

    • Sampler: Euler a - Karras/SGM Uniform

    • Resolution: 832 x 1216, 1024 x 1024

    • CFG: 5.5

    • Sampler: DPM++ 2M SDE - Karras

    • Resolution: 832 x 1216, 1024 x 1024

    • CFG: 6.5

    Quality tags:

    Pony

    score_9, score_8_up, score_7_up, (masterpiece:1.2), (best quality:1.3)

    Illustrious

    (masterpiece), (best quality), masterwork, semi-realistic

    Negative prompt:

    Pony

    low quality, score_6_down, score_5_down, score_4_down, score_3_down, score_2_down, score_1, NEGATIVE_HANDS

    Illustrious

    low quality, 2koma, 4koma, signature, watermark, bad anatomy, jpeg artifacts, signature, watermark

    Positive lighting tags:

    low-key lighting, lo-fi, glow, dynamic cinematic lighting, ray_tracing, global illumination

    Negative lighting prompts:

    bright, spotlight, front lit, camera flash

    I am not sure how those tags should affect the image. lol they are just what I use.

    Optional: ADetailer for face details

    Description

    • Sampler: DPM++ 2M SDE - Karras

    • Resolution: 832 x 1216, 1024 x 1024

    • CFG: 6.5

    Model merged with SILVER MAGIC XL by RAMTHRUST

    FAQ

    Comments (3)

    hushichoNov 4, 2024· 4 reactions
    CivitAI

    I wish this weren't the final! It's a promising checkpoint, but it doesn't seem to have a lot of flexibility, and I can't get anywhere close to the output you're getting. Always glad to see uploads from you though!

    Vauxz
    Author
    Nov 4, 2024· 2 reactions

    Thanks for your feedback! I am able to work on it on my free time as I am still not 100% satisfied with this latest model. I want to improve prompt adherence and improve dark and low exposure images. The future releases would have different version names instead of numbers so as not to be confusing. lol

    hushichoNov 4, 2024· 2 reactions

    @Vauxz I look forward to trying these new versions and I'm really glad you're going to continue developing it, it has a lot of really great qualities!

    Checkpoint
    SDXL 1.0

    Details

    Downloads
    489
    Platform
    CivitAI
    Platform Status
    Available
    Created
    11/2/2024
    Updated
    5/12/2026
    Deleted
    -

    Files

    vauxzAnimeMergeIllu_xlV11.safetensors

    Mirrors