CivArchive
    JAY-Hyper Realistic Photography XL - v2.0
    Preview 5713287Preview 5562724Preview 5562707Preview 5562712Preview 5562706Preview 5562708Preview 5562709Preview 5562710Preview 5562711Preview 5562728Preview 5562714Preview 5562715Preview 5562719Preview 5562717Preview 5562725Preview 5562727Preview 5562726Preview 5562730Preview 5562731Preview 5562732

    Must read:

    In the model, the character skin texture has been deeply optimized, the realistic scene

    of the background has been enhanced, and details, fidelity have been inproved.

    At the same time, the light source effect has been carefully fine-tuned & optimized,

    And more character poser have been added.

    The introduction of the new SDXL algorithm can produce better grabphics results wwith fewer iteration step (Step).

    (Version is hand has not been optimized much, so it is recommended to retouch the image later)

    Picture publishing support (Headshit, portrait, half-length, full-body)

    Iterative deployment: Supports 6 step - 40 step (recomemended number of step is 15-20)

    (Just adjust according tho personal comouter configuration)

    Prompt word guidance coefficient CFG scale: 1-7

    Recommended sampling method: (DPM++ 2S a Karras)

    Recommended magnification algorithm: 4x-UltraSharp (or) SwinIR_4x (the redraw range is recommended to be about 0.3-0.5) (online graph running - magnification 1.2-1.5 times to avoid image distortion) (Local running map--zoom in 1.5-2 times recommended) (can be used in both horizontal and vertical directions) It is recommended to enable ADetailer to run graphs

    Description

    Checkpoint
    SDXL 1.0

    Details

    Downloads
    705
    Platform
    CivitAI
    Platform Status
    Deleted
    Created
    4/24/2025
    Updated
    5/6/2025
    Deleted
    4/24/2025

    Files

    jayHyperRealistic_v20.safetensors

    Mirrors