CivArchive
    OIA Dreamweaver - 0.10
    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


    Dreamweaver focuses on the following aspects during testing:

    • Model's portrayal of light and shadow;

    • Diversity in facial features, face shapes, and details such as clothing, hairstyles, hands, etc.;

    • Portrayal of large scenes;

    • Model's color, composition, and the unity between subject and background;

    • Model's compatibility with LoRA;

    • Model's response to different visual style cues. You can use cues like illustrator, Anime, Comic, Watercolor, Oil Painting, Photographic, Digital Art, Fantasy Art, Neon Punk, Low Poly, Line Art, 3D Model, Pixel Art to test its response.

    • Model's portrayal of mechanical, mecha, animals, and plants elements;


    Anyway, Dreamweaver is absolutely, absolutely, absolutely worth collecting and mastering (important things need to be said three times, haha)! If you are a talent, she will be the brush in your hand, effortlessly turning your imaginative ideas into reality. Looking forward to everyone's feedback on mastering her, her amazing expressive power will not disappoint you.


    Let's create the world together! Big big hugs to all the students who love Dreamweaver! Love you all!


    Recommended settings:

    • CFG Scale: 3-9 (7 is the most stable and conservative value, below 7 leans towards realism, while higher values approach the cues for more precision)

    • Clip skip: 2

    • Sampler: Restart, DPM++2M Karras (Restart is preferred)

    • Sampling steps: 25-30 (very high values yield minimal gains and waste time)

    • Hires step: 4x-UltraSharp (high-resolution restoration is generally unnecessary)

    • VAE: built-in


    Regarding versions: the default version is based on SD1.5, and versions with the suffix XL are based on SDXL1.0.

    Both 0.10 and 0.10XL versions yield explosive results, competing with any large model and possessing stronger cross-field capabilities. The sample images provided are all direct outputs without LoRA.


    Suggested Writing Style for Cues:

    1. Photography / Photorealistic Style

    Cue: photographic of <your cue>. BREAK, 35mm photograph, grainy, professional, 4k, highly detailed, bokeh

    Reverse Cue: drawing, painting, crayon, sketch, graphite, impressionist, noisy, blurry, soft, deformed, ugly

    Example: photorealistic of a woman sitting at a cafe. BREAK, 35mm photograph, grainy, professional, 8k, highly detailed, Leica M50 F/2.8

    ... (remaining content remains unaffected) ...

    FAQ

    Details

    Downloads
    2,006
    Platform
    ShakkerAI
    Platform Status
    Available
    Created
    12/29/2023
    Updated
    9/6/2024
    Deleted
    -

    Files

    OIA DreamWeaver_0.10.safetensors

    Mirrors