stable-diffusion text-to-image

Intro

Been using this mix for a week. I'm a noobie. </br></br></br>

Recommends:


<b>WxH</b>: Multiples of 128 (i.e 512x512)

<b>VAE</b>: vae-ft-mse-840000-ema-pruned.ckpt

<b>Hypernetwork</b>: anything-v3.0.vae.pt (optional)

<b>HiRes Fix</b>: Recommended </br></br></br>

Suggestions:


<b>anime results</b>:

sketch, watercolor, illustration

<b>realistic results</b>:

Realistic, role-playing, face, portrait, painting of a beautiful rpg champion

<b>negative prompts</b>:

(that yields consistent results)

(disfigured), (bad art), (deformed), (poorly drawn), (extra limbs), (close up), strange colors, blurry, boring, signature, letters, watermark, out of frame , worst quality , text , blurred , monstrous , hideous , ugly , duplicate , cropped , mutilated , horrifying

</br>

Workflow


Step Interpolation Method Primary Model Secondary model Tertiary Model Merge Name
1 Weighted Sum @ 0.4 RPG AnythingV3.0 n/a AnyRPG-mix
2 Add Difference @ 1 AnyRPG-mix Baka-Diffusion V1 berrymix-v3 BakaBerry-d1
3 Weighted Sum @ 0.25 BakaBerry-d1 r34_e4 n/a Syx0 Blend V3

Socials:</br> https://github.com/axsddlr

https://www.reddit.com/user/ayyser </br></br></br>