CivArchive
    Bouncing penis Concept - V2.1
    NSFW
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined
    Preview undefined

    Link to TensorArt: Bouncing penis Concept - v2.1 - Free AI Model | TensorHub Art | TensorHub Art

    V2.1: I uploaded a much earlier epoch of the same training as V2. Compared to V2, this epoch doesn't seem to break the faces as much. Different from 2 earlier versions, all sample images of this epoch didn't go through Adetailer to fix faces and just ran as it is. Same recommended prompt as V2.

    V2: offering better quality and stability in generating afterimage of penis while also having increased chance of getting bar censored

    KNOWN ISSUE: FACES BROKEN if not using face fixes or Adetailer

    Recommended strength: 0.7-0.8

    Prompt: bouncing penis, afterimage + uncensored

    V1: My 1st concept Lora. It's not the best.

    Sometimes the afterimage of the penis looks subtle, with only a few motion lines and it might not be to your desires or sometimes it doesn't work.

    I'm not sure what the best prompt might be but you could try: (bouncing penis, afterimage:1.2) + other penis-related tags

    Description

    FAQ

    Comments (1)

    ThatNewDoodJan 17, 2025· 1 reaction
    CivitAI

    V2.1 is sooo much better!!

    I'm happy to see that v2.1 got such a large improvement from v2.0, a lot of my annoyances with v2.0 are heavily reduced, and the faces being altered is from what I noticed completely fixed!

    I can now happily recommend this Lora to people.

    So good job @InVain1109.

    Certain issues do still persist (though way less potent).
    This Lora does make it imagine things which are not in the prompt. e.g. ponytail, hairband, demon horns.
    But they're less common now.

    I do not know how you train your Lora, but maybe being more specific (e.g. having a lot of tags for specific hairstyles and stuff like that) in your training prompt will improve that.

    (Do not quote me on that, I do not have much experience making Lora's myself)

    I am surprised to see in the changelog that you ran the training data for V2 through ADetailer.
    While ADetailer is an amazing tool for generation, in my experience, you should never ever alter training data.
    The only exception to this rule is removing text bubble's with the speech bubble remover.

    LORA
    Illustrious

    Details

    Downloads
    1,378
    Platform
    CivitAI
    Platform Status
    Available
    Created
    1/17/2025
    Updated
    6/12/2026
    Deleted
    -
    Trigger Words:
    bouncing penis, afterimage
    motion lines
    uncensored

    Files

    BouncingPenis_v2.1.safetensors

    Mirrors

    Available On (1 platform)

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