This model is based on anything4.5 trained from the picture of lillly, and it is recommended to use weights between 0.4 and 1.0. There have been many controversies about the training and use of art style models, but they are also a difficult part to avoid in the process of AI painting progress. Therefore, it is recommended to use such art style models as a kind of existence similar to potions, by matching Loras of different art styles with different weights to form images with different styles (although this may lead to bad results, it is worth trying).Alternatively, it can be used to match your favorite character models. Anyway, I hope not to compare it with the artist's own work and make rude remarks, which will harm the community's environment. At the same time, there are still many problems to be discovered in this model. In short, hope you have a good time.

lillly Style

This version has a lower loss value (loss=0.112) and has been tested on anything 4.5 with the effect shown in the example figure

Files with SHA256: 03c810e428454155db6b618d6376f5ebe3d1c2021d1102a8bcab1dcaf9f76a07