CivArchive
    Qwen Image Female Anatomy Fix — NSFW LoRA Triggerless v2 - Triggerless v2
    NSFW
    Preview 141547069
    Preview 141547040
    Preview 141547044
    Preview 141547053
    Preview 141547062

    An anatomy-assist LoRA for Qwen Image 2512, designed to improve coherent adult female external anatomy in otherwise strong generations.

    This is not a standalone character or style model. It works best beside another adult-oriented LoRA that supplies the person, scene, identity, or visual language.

    Usage

    • Trigger: none — use natural language

    • Suggested weight: 0.4–0.8

    • Best use: anatomy correction and refinement

    • Training: 1,046 images, rank 32, batch size 8, 7,600 steps

    Limitations

    Higher weights may overfit or overpower another LoRA. It cannot repair impossible poses, severe occlusion, or fundamentally broken body geometry. Test multiple seeds, views, distances, and body types.

    For lawful depictions of consenting adults only. Do not use with minors, non-consensual scenarios, or identifiable real people without permission.

    Full documentation, reference scripts, and governing license:

    https://huggingface.co/perpetual3x/Qwen-Image-Female-Anatomy-Fix-NSFW-LoRA-Triggerless-v2

    License notice: This upload remains governed by the custom non-commercial research license in the linked Hugging Face repository. Any platform-generated license label does not replace those terms.

    Platform permission: Civitai may host this model and provide generation-service use under its platform terms. This does not grant users permission to sell the model or its merges.

    Description

    Initial Civitai release of the triggerless anatomy-assist edition. Natural-language prompting; recommended weight 0.4–0.8. Trained for Qwen Image 2512.

    LORA
    Qwen

    Details

    Downloads
    80
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/2/2026
    Updated
    9/2/2026
    Deleted
    -

    Files

    anusv2.safetensors

    Mirrors

    CivitAI (1 mirrors)