Type | |
Stats | 9,248 6,819 |
Reviews | (1,977) |
Published | Jan 25, 2023 |
Base Model | |
Trigger Words | artbytokiame |
Hash | AutoV2 9FBDAF69EC |
LORA trained on art by Tokiame, particularly the art pieces with thick outlines, it was trained on Anything 4.5, the best results i had were on Anything 4.5. i think a weight of around 0.85 was pretty good.
I tried it on EerieOrangeMix2 and personally i found the results weren't as good, and the weight needs to be higher (around 1), but i didn't experiment much, tell me how it goes for you and please do share your creations.
it works best on people and portraits, but can have some cool results with landscapes as well. and sometimes can give some interesting results to none-anime generations/models.
some of the prompts in the examples aren't mine.
trigger word is: artbytokiame
edit: I added a few more examples i made using protogen infinity, the results look really good, the post apocalyptic ones at the end were made using this model.
also a tip I found, if you want to go crazy on the thick dark outlines, use darkest diffusion TI embedding together with this LORA, the results of the combination are really cool, but look slightly less anime