does it work with anime? any anime images were in the dataset?

#6
by krigeta - opened

Hi, when you were training this lora, did you use any anime images as well?

Hi, when you were training this lora, did you use any anime images as well?

no

hey @dx8152 are you planning to implement it?

hey @dx8152 are you planning to implement it?

Please test whether it works first. If it doesn't, I will consider training specifically for animation later.

Yeah, it works. but remember anime characters have weird mouths which are drawn for specific angles, but for the scenery it works fine. Thanks for this lora!
photo-collage.png

There, another test. I honestly find nothing wrong with it. even the mouth seems fine.
photo-collage.png (2)

There, another test. I honestly find nothing wrong with it. even the mouth seems fine.
photo-collage.png (2)

Because the Qwen model is very powerful, its generalization ability is excellent, even without specific training on animation.

Hey @Kinduun thank you for the testing. I will try it asap on cloud, need some time to add up some money for that though, then I will share the results.

Sign up or log in to comment