This model is trained on the basis of the model animefull-final-pruned(<a target="_blank" rel="ugc" href="https://huggingface.co/a1079602570/animefull-final-pruned/tree/main">https://huggingface.co/a1079602570/animefull-final-pruned/tree/main</a>).A total of 65 pictures of a illustrator oyuwari_u[おゆわり](<a target="_blank" rel="ugc" href="https://www.pixiv.net/users/44898022">https://www.pixiv.net/users/44898022</a>) I like are used in the training.The training is completed on the cloud platform using 3090.You can try a model weight of 0.5-0.8 and a cfg value of about 7 and I recommended to set the cilp skip value to 1.The lora model has good application effect on some anime models, but it has limited effect on some 2.5d models, such as AbyssOrangeMix2. I haven't done many model tests. You can test the model effect on your own model.Still trying to improve the model, maybe.Finally, I wish you all a good time.

Files with SHA256: 671b461207578eabaf85491a6f3d8b3df1d8c301ea80cce4abaa32c9e27a35d2