CivArchive
    [PonyXL] More Saturation | Make the picture more realistic and gradient | SP - v1.0
    NSFW
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined

    The goal is to make the picture more saturation without affecting the style of the original Checkpoint.

    Make the performance of light and shadow in the picture more realistic and gradient.

    Just like the saturation of the photo. Let the picture have a story like the photo.

    Realistic Checkpoints, the weight is recommended to be 0.5 ~ 1,

    Semi-Realistic and 2.5D Checkpoints, the weight is recommended to be 0.7 ~ 1

    2D and Toons Checkpoints, weight recommendation 1 ~ 1.2

    Known issue: Using these prompts: "cinematic, film, still" will not keep the style consistent.

    Thanks everyone for your suggestions, special thanks to LindezaBlue.

    Example:

    Note1: Weight 0.7 is one of the best choices, as it can enhance details without affecting the selected Checkpoint too much.

    Note2: You will have the best experience with CosplaySerendipityPNY.

    Note3: I have tested many Pony Checkpoints so far and they all performed well.

    But it doesn’t mean that this Lora is suitable for every kind of Pony Checkpoint.

    Description

    First Verssioon

    FAQ

    Comments (4)

    120458Aug 5, 2024· 1 reaction
    CivitAI

    Interesting concept, however even with seeds clamped and other settings locked it still changes the checkpoint style. This lora isn't good for consistency.

    springshsiao350
    Author
    Aug 5, 2024· 1 reaction

    I think you're right in bringing up that this reduces consistency. But thinking about it from another perspective, this also makes the original Checkpoint more flexible.

    Thank you

    120458Aug 6, 2024

    @springshsiao350 I guess.. it's an odd thing to use a lora to make a checkpoint more "flexible." I think my checkpoint and many others are flexible in their own right, however introducing more variation despite having a seed to lock in an image I like and not getting consistent results and when the lora is removed and results are back to being consistent to the style I want, then the issue isn't the lora is "flexible" it's that it overwrites the style that was purposely trained into the checkpoint, and that is NOT a desirable thing to have.

    springshsiao350
    Author
    Aug 6, 2024· 2 reactions

    @LindezaBlue 
    You're right.
    I found that I was thinking about the operation of Lora from the perspective of an engineer rather than a "painter". I'm currently trying to improve in the direction you suggested. You gave me very good advice, thank you.

    LORA
    Pony

    Details

    Downloads
    219
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/3/2024
    Updated
    5/1/2026
    Deleted
    -

    Files

    SP_Saturation.safetensors

    Mirrors

    Huggingface (1 mirrors)
    CivitAI (1 mirrors)

    Available On (1 platform)

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