This LoRA is trained on screenshots from Arcane. It features many League of Legends characters, namely Jinx, Vi, Caitlyn, Ekko, Jayce, Mel, Sevika, Silco, Vander, and Victor. This model can recreate both the style and characters from the Arcane series. Trained on Pony Diffusion V6 XL, it possesses Pony's ability.
Update:
add Flux version which support 4 characters which are Jinx Vi Mel Caitlyn.
Usage Tips:
Use with Pony Diffusion or a Pony-based model.
Use with a pony prompt.
Trigger words for characters (some characters are tagged but not enough data):
main female:
jinx,vi,caitlyn,mel,sevika,main male:
silco,jayce,viktor,ekko,vanderorphan:
powder,yung vi,yung ekko,mylo,claggorKDA:
ahri kda,akali kda,evelynn kda,kaisa kdaimagine dragons (not good):
ben mckee,dan reynolds,daniel platzman,wayne sermon,jidundergound people (not good):
deckard,dustin,mek,ran,thieram,finnothers:
ambessa medarda,elora,grayson,heimerdinger,hoskel,cassandra kiramman,howl,marcus,skyyoung
You can also add LoRA for mimicking the Arcane style.
Description
Updated v2.0, Train with images from 2nd season of Arcane.
More character and outfit from Season 2.
FAQ
Comments (17)
Buddy, Grayson is female, not male.
lol, my prompt is automatically created. Sometime, thing like this happen.
@titansteng All good haha
v2 version creates a lot of chromatic aberration and characters look kinda deformed for some reason, previous versions didn't have this issue
It might be the compression of the dataset. You can see from the size of training data. v2 have a lot more image but the size is also smaller.
Initially, you can mix with previous version. I will come back and try to fix it, if I can find some time.
I just updated a newer version of v2.0. I hope this help.
Are you using Hires. fix in the Stable Diffusion Web UI when generating the example images? If I don't use that, the images are very low quality. If I try to use the Hires. fix the images are better quality but they end up looking very grainy
No. I didn't use that.
check out the newer version of v2.0. I just updated it. a bit improve in quality.
I also use v2. Which base model did you use?
@birdupthemusical506 I use basic pony XL model, and some pony-base that I merged for myself. I have reuploaded the newer version of v2.0 to fix the quality of the previous v2.0 even on the basic pony model. I hope that also fix the problem that you mention.
Please, add Flux version for all the other chars, please <3
Great work!
I use FLux 1d
and add this to the lora files
it crashed down
Illustrious version please!
done!
@titansteng Thank youuuuuuuuu ♥















