Type | |
Stats | 1,161 |
Reviews | (151) |
Published | Jul 15, 2024 |
Base Model | |
Training | Epochs: 1 |
Usage Tips | Clip Skip: 2 |
Hash | AutoV2 EAC1620586 |
This is a model merge of Hadrian DeliceXL 2.0 with two of my LoRAs, MeMaXL V3 Hadrian and Cool Rance XL V2 Hadrian.
The goal of this model is to product high quality 2d flat anime and backgrounds.
Why use this mix?
It can create true 2d anime in a flat/anime screencap style
It can produce illustration grade backgrounds
It's pony based so it can do a great deal of nsfw
It's extremely proficient at "1girl"ing
Where does it suck?
The lora compatibility is extremely hit or miss, just assume it won't work well, you'll have better luck with character loras. Styles are mostly miss. I haven't tested a lot of concepts.
How to use?
Start your prompt with the text below and use Danbooru tags to describe your image. For default style.
Prompt:
score_9,score_8_up,score_7_up,score_6_up,source_anime,
For a more 2d screenshot style.
Prompt:
score_9,source_anime,anime screencap,
Negatives:
mosaic censoring,bar censor,censored,source_cartoon,source_pony,source_furry,simple background,
Sample:DPM++ 2S a
Steps: 26
CFG Scale: 7
Extension: [sd-webui-color-enhance] set to 0.5 (this increases the saturation a bit)
Upscaling: Follow this method
I like the results but I don't wish to use your model, what do?
You can just use Hadrian DeliceXL 2.0 with the two loras mentioned above (both at 0.5 strength) and get pretty much the same results. If you prefer to use AutismMix or base Pony, you can do the same using the AutismMix version of the loras.
A special thanks to Hadrian for making such a great model.