|
--- |
|
language: |
|
- en |
|
license: creativeml-openrail-m |
|
tags: |
|
- stable-diffusion |
|
- stable-diffusion-diffusers |
|
- text-to-image |
|
- diffusers |
|
- safetensors |
|
inference: false |
|
--- |
|
|
|
Had these for a while, made just as an experiment. I'm only uploading them as an online backup, I don't claim anything except that I'm surprised they output anything coherent at all. Recommend using High res fix at higher resolutions, so the second pass fixes and adds extra detail. |
|
|
|
### ABSOLUTE MADMAN mix |
|
|
|
A mix of too many models, some which were already mixes on their own. Most were merged on a 0.3 to 0.4 ratio, it was just a dumb experiment tbh, no real science at all. |
|
The only notable quirk is that it does traditional media weirdly well. |
|
|
|
Models merged include samdoesbimbosmix + anything + eimis + elysium + trinart + hassan + cafe + 852 + my two models from my art + popn model + 4 additional unpublished models i trained separately as an experiment + i forgot they were too many. |
|
|
|
Examples: |
|
|
|
 |
|
|
|
Using some prompts from: https://rentry.org/54d9o |
|
|
|
### FIXER mix |
|
|
|
Only called that because it worked well at fixing my own art, it's not only meant for fixing, that's just a dumb name. Regular outputs are ok enough. It has about half the models used + HD + EB |
|
|
|
 |
|
|
|
### wheatpopman mix |
|
|
|
A mix of istolemyownwheat, popn and an experimental model trained on bleedman (lol). It's alright for thick lineart. |
|
|
|
 |
|
|
|
|
|
|
