MISA / README.md
ThePioneer's picture
Update README.md
caef5d2
metadata
license: creativeml-openrail-m
language:
  - en
  - ja
tags:
  - art
  - safetensors
  - stable-diffusion
  - text-to-image
thumbnail: https://huggingface.co/ThePioneer/MISA/resolve/main/model_face.png

merged_sample

Download

About

Mix of
Impressive
SD
Art models

MISA. A powerful wide range model based on Waifu Diffusion 1.5 Beta 2 Aesthetic, Quattro4Merge+i, and Illuminati V1.1.

The most interesting thing about this model is that it is pretty good at generating half Asian half European beauties.

The biggest improvement is that we no longer need long negative prompts. In most cases, it will work fine with just a couple of triggered negative tags.

Samples

See the civtai page for sample prompts except for the last one that was blocked.

sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images sample_images

Extra (only on Huggingface!)

sample_images

(one chibi girl:1.2), (best quality), (white background:1.3), full body, the girl has (bright light blue hair long hair), [[blue eyes]], the girl wearing (short sleeves, short navy blue gloves, angle navy blue beret with long ribbons), (laced navy blue coat, laced and short navy blue capelet:1.2), lolita fashion, (white background:1.3), (one chibi girl:1.2), Negative prompt: nfixer, nfixernext, re-badprompt,badquality, (worst quality, low quality:1.2), Steps: 35, Sampler: DPM++ 2M Karras, CFG scale: 11, Seed: 919994676, Size: 1024x1024, Model hash: f0f7b010f4, Model: misa