CivArchive
    XIN Realistic Illustration Thick painted Half body Portrait - v1
    Preview 57756947Preview 57756953Preview 57756952Preview 57756951Preview 57756957Preview 57756955Preview 57756950Preview 57756948Preview 57756944

    This model can transform the originally realistic character images into a two-dimensional illustration style with thick texture, strong light and shadow effects, and messy hair. Specifically, it will enhance the pen touch and color contrast of the image, making the characters look more like hand drawn illustrations rather than purely real photos. This style adds artistic illustration texture and pursues personalization on the basis of realism, which is very practical.

    Description

    This model can transform the originally realistic character images into a two-dimensional illustration style with thick texture, strong light and shadow effects, and messy hair. Specifically, it will enhance the pen touch and color contrast of the image, making the characters look more like hand drawn illustrations rather than purely real photos. This style adds artistic illustration texture and pursues personalization on the basis of realism, which is very practical.

    Sampling and iteration parameters

    Iteration steps: forty

    Sampling method: DPM++2M Karras

    High Resolution Enhancement

    Magnification algorithm: Latent

    High score iteration steps: one hundred and twenty

    Redraw amplitude: 0.4 or 0.5

    Magnification factor: two

    Other parameters

    Refiner strength: zero point eight

    Generate size: 512 x 768

    Guidance coefficient: seven

    Seed value: one billion seven hundred and seventy-four million two hundred and sixty-five thousand six hundred and six

    ADetailer configuration

    Facial details: mediapipe_face_mesh

    Hand details: hand_yolov8n.pt

    Lora sets weights: zero point eight

    Optimization suggestions

    Parameter fine-tuning: Adjust the iteration steps and guidance coefficients appropriately based on the generated effect to ensure rich image details.

    Local refinement: If necessary, other high-precision facial or hand refinement models can be combined to further enhance the representation of local details.

    Lora weight: The weight can be adjusted by 0.2 based on the generated effect to achieve the best style performance.

    VAE:chilloutmix_NiPrunedFp32Fix.vae.ckpt

    LORA
    SD 1.5

    Details

    Downloads
    70
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/14/2025
    Updated
    11/12/2025
    Deleted
    -
    Trigger Words:
    lingluanhoutu

    Files

    XINlora_lingluanhoutu.safetensors

    Mirrors