Trained with captioned images for 30 epochs










LoRA is trying to make a vibe of 90's and early 00's, also mixed with dreamcore, kidcore aesthetics. I'm not sure why, but I associate that vibe with modern fusion jazz musicians: KNOWER, Louis Cole, Nelward, Sam greenfield, Bill Wurtz, so on. And that's what i've tryed to recreate. I think it came out well :)
Sometimes model tends to "deform" stuff, if you see this, try to decrease LoRA's weight and find the sweet spot between deformated results and stylization. I found that weights between 0.5 and 1.5 works best. The shorter the prompt, the less weight you need. And in opposite, as the prompt grows, you should try to increase weight up to 1.5
You can support me and request new lora's/checkpoints at: https://boosty.to/smgbuddies/donate
Trained with captioned images for 30 epochs
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
