Initial release of LoRA trained on 169 images from fancaps.net.










Model used: AbyssOrangeMix3
Recommended VAE: orangemix or kl-f8
Recommended LoRA weight: 0.8-1 (I used 1 for most previews)
Trigger words: genos, anime, one punch
Recommended words in prompt: 1boy, cyborg, colored sclera
Negative prompt example:
(worst quality, low quality:1.4), 3d, greyscale, monochrome, textOn all images from preview I used DPM++ 2M SDE sampler, but you can use what you like.
All other settings the same as for AOM model:
Steps:
DPM++ SDE Karras: Test: 12~ ,illustration: 20~
DPM++ 2M Karras: Test: 20~ ,illustration: 28~
Clipskip: 1 or 2
Upscaler :
Detailed illust → Latenet (nearest-exact)
Denoise strength: 0.5 (0.5~0.6)
Simple upscale: Swin IR, ESRGAN, Remacri etc…
Denoise strength: Can be set low. (0.35~0.6)
This LoRA was trained on 169 images which I scraped from fancaps.net (I made tampermonkey extension which adds a download button under each image). I used wd14 tagger extension for Stable diffusion to automatically tag my images and LoRA trainer colab for training (169 images with 3 repeats, 2 batch size, 10 epochs).
I plan to do a lot of LoRAs based on the One Punch Man anime. If you are interested, you can follow me.
Initial release of LoRA trained on 169 images from fancaps.net.
1. 재게시된 모델의 권리는 원 제작자에게 있습니다.
2. 모델 원작자가 모델을 인증받으려면 공식 채널을 통해 SeaArt.AI 직원에게 문의하세요. 저희는 모든 창작자의 권리를 보호하기 위해 노력합니다. 인증하러 이동
