CivArchive
    Dream Shattered | XL Realistic Big Model | NeverDream - Dream Shatter-V5.6
    Preview 1
    Preview 2
    Preview 3
    Preview 4
    Preview 5
    Preview 6
    Preview 7
    Preview 8
    Preview 9
    Preview 10
    Preview 11
    Preview 12
    Preview 13
    Preview 14
    Preview 15
    Preview 16
    Preview 17
    Preview 18
    Preview 19
    Preview 20

    Description

    Model introduction:

    Welcome to the Dream Shatter model. This model is a general-purpose model with a real style. It uses more than 5,000 realistic style photos, more than 2,000 ancient Chinese pictures, and more than 2,000 Western fantasy style pictures. A total of 10,000 images completed the initial training, and 1,000 images were selected on this basis The model was fine-tuned through several rounds of iterations. The characters generated by this model are far superior to the SDXLbase model in terms of texture and detail, and also inherit the high quality lighting and shadow performance of the SDXL Base. Since this model uses more fantasy style images, it will be more expressive in fantasy style. Everyone is welcome to try it out

    and return the picture.

    In this model, you can not only experience the unique charm of Guofeng, but also experience the wonderful world of Western magic. At the same time, you can also feel the thrilling and scary atmosphere that touches the soul.

    Generation suggestions: 1. Recommended number of sampling steps:

    30-60 (if possible, you can use a higher number of sampling steps. The higher the number of sampling steps, the richer the details, limited to local conditions, not tested by the author) 2. Recommended sampler (

    Key points! Focus!)

    DPM ++ SDE Karras (recommended) DPM2 a Karras (recommended)

    DPM ++ 2M

    SDE Karras 4, resolution (focus!)

    SDXL does not support 512*512. Please refer to the SDXL print resolution requirements. We recommend drawing in sizes such as 768*1024, 1024*1024, 1024*1280, 768*1280, etc. If you exceed the size, the body ratio may be out of proportion.

    5. Positive hint words

    : You can add quality words to get better results

    Recommended

    style words 90s HongKong (

    90s Hong Kong style) cinema photo (movie photo)

    Horror film (horror style)

    Vintage Polaroid (retro film)

    Street Documentary (street

    shot)

    Retro Sci-Fi (retro sci-fi)

    Film Noir (black and white film)

    The author uses negative cues: drawing, painting, crayon, sketch, graphite

    , impressionist, noisy, blurry, soft, deformed, ugly, bad-hands 5, (worst quality, low quality: 1.4), ng_deepnegative_v1_75t, easynegative, bad-artist, bad face, bad anatomy, bad perception , multiple views, mutated hands and fingers, interlocked fingers, twisted fingers, excessively bent fingers, more than five fingers, lowres, bad hands, text, error, missing fingers, extra digit, difficult digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry, artist name, low quality lowres multiple breasts, low quality lowres mutated hands and fingers, more than two arms, more than two hands, more than two legs, more than two feet,

    6. Others: The Refine model is essential. You can get good results when this model comes straight out. Also, Asian people don't recommend using refinement. Please use Adetailer for facial restoration

    .


    FAQ

    Details

    Downloads
    45,839
    Platform
    ShakkerAI
    Platform Status
    Available
    Created
    9/6/2023
    Updated
    3/17/2025
    Deleted
    -

    Files

    NeverDREAMXL_V5.6.safetensors

    Mirrors

    HuggingFace (1 mirrors)
    ShakkerAI (1 mirrors)