CivArchive
    Mewto Build - ZImageBase and ZImageTurbo - v2.0
    NSFW
    Preview 119287737
    Preview 119287739
    Preview 119287740
    Preview 119287743
    Preview 119287738
    Preview 119287741
    Preview 119287742
    Preview 119287735
    Preview 119287734

    For producing women with a "mewto build", aka curvy woman. Helpful prompts for success: "wide hips", "round belly", "curvy woman", "curvy hips".

    Description

    Trained on Z-Image Base. Enhance results for Z-Image Turbo. I get okay prompt adherence with Base. I think its more so dialing in your prompt then anything else.

    FAQ

    Comments (9)

    TypicalLoraLoverJan 30, 2026· 2 reactions
    CivitAI

    how is your prompting so good? I'm not good at prompting myself and i don't know a good way to generate prompts...

    TheAINovice
    Author
    Jan 30, 2026· 4 reactions

    The best strat i’ve found is use a photo that is similar to your desired image and run it through an LLM. After you do that you can tweak the prompt it spits out however you like. I use Qwen 3vl. You can run it entirely through comfyui with some custom nodes or run it through LM studio. I’ve found that LM studio is a lot faster though. Shoot me a message if you want me to give more detailed instructions.

    TypicalLoraLoverJan 30, 2026· 1 reaction

    @TheAINovice Ah thanks for that info! if i need any additional help, i will send you a message for sure!

    tenhunter500Jan 30, 2026· 1 reaction

    @TheAINovice just a passerby but thank you so much I followed your strat and wow my images and video come out so much better with the LM prompting. Thanks a lot!

    BinaryBottleBakeJan 31, 2026

    @TheAINovice Which Qwem 3vl model do you use, and what kinda of prompt do you use for it to help write out prompts?

    TheAINovice
    Author
    Feb 1, 2026

    @BinaryBottleBake I use huihui-ai/Huihui-Qwen3-VL-8B-Instruct-abliterated

    TheAINovice
    Author
    Feb 1, 2026· 2 reactions

    @BinaryBottleBake Its best to use prompts that fit what you want but I've been using this: "Analyze the provided image carefully and generate a long. highly detailed description suitable for a high quality realistic text-to-image model.

    Describe the subject's appearance, clothing, body proportions, pose, facial expression, and mood. Make sure to emphasize sensual or provactive elements if present.

    Include detailed descriptions of:

    Hairstyle, hair color, and skin texture

    Clothing materials, fit, and how they interact with the body

    Visible tatoos, accessories, or defining features.

    Body posture and how it accentuates the form of the subject

    Environment, background elements, lighting sources, and color palette

    Cameral angle, framing, depth of field, and photographic style

    Write in a cinematic erotic-glamour tone that feels confident, polished, and visually rich.

    Output the result as three paragraphs. The first paragraph focuses on the subjects appearance and pose, the second paragraph focuses on the clothing, the third paragraph focuses on the environment and photographic style.

    Avoid mentioning watermarks, and only output the description."

    NIKULINAFeb 12, 2026
    CivitAI

    can you train forum.candidgirls.io lora?

    TheAINovice
    Author
    Feb 12, 2026· 1 reaction

    Will do 🫡

    LORA
    ZImageBase

    Details

    Downloads
    766
    Platform
    CivitAI
    Platform Status
    Available
    Created
    1/30/2026
    Updated
    5/13/2026
    Deleted
    -

    Files

    MewtoV2.safetensors

    Mirrors

    CivitAI (1 mirrors)