MagicalMix_v2 / README.md
mekabu's picture
modified README.md
8154b4b
metadata
license: creativeml-openrail-m
pipeline_tag: text-to-image
tags:
  - stable-diffusion
  - stable-diffusion-diffusers
  - text-to-image
library_name: diffusers

What's this model

Based on MagicalMix v1, we aimed for a softer picture.

By samplers, both anime and softer (like illustration) styles can be expressed.

Setting and examples

Embeddings

EasyNegative

bad_prompt_version2

VAE:pastel-waifu-diffusion.vae.pt

[Softer]

Sampler: Euler a
Steps: 40
Hires fix: x2
Upscaler: Latent
Hires steps: 30
Denoising strength: 0.5
CFG scale: 8
Clip skip: 2

[Anime]

Sampler: DPM++ 2M Karras
Steps: 25
Hires fix: x2
Upscaler: R-ESRGAN 4x++ Anime6B
Hires steps: 20
Denoising strength: 0.5
CFG scale: 8
Clip skip: 2

**Any other good presets out there, let me know!**

Model recipe

No. Interpolation Method Model_A Model_B Tertiary Model Output
1 Weighted sum @ 0.45 Basil_mix_fixed uenomigiDB_v1 - Bronze1
2 Weighted sum @ 0.7 Bronze1 ligneClaireAnime_v1 - Bronze2
3 Weighted sum @ 0.6 Bronze2 abysshell_ver3 - Bronze3
4 Add Difference @ 0.75 Bronze3 dalcefoHitoria_v3 EasyNegative Silver
Model_A Model_B Weight Output
MagicalMix_v1 Silver SMOOTHSTEP MagicalMix_v2