(Don't forget to share your creations! :D) Made with 30 images, average image size: 300x500, still good results. Recommended between 0.8 and 1 network multiplication Trained on AbyssOrangeMix2-nsfw (I used negative prompts, selected this model because I like it because I had problems with Anything v3 results) It seems it can get sometimes wrong the order of things (I asked it a pink shirt and blue shorts, sometimes it confused the color of the shirt with the color of the shorts), but most of the time does thing right I'm planning to do some others characters of the series, if you want I prioritize one, ask me and I will see if I can
|