The original intention of this model is to create a fantasy style while enriching the details of the picture. After testing, it adapts well to 2D and 2.5D large models, but there are still many problems. It will continue to be updated and optimized in the future.

The weight of lora is between 0.6-1. The smaller the weight, the fewer details in the picture and the cleaner the character image. The larger the weight, the richer the picture and the more details, but at the same time it will produce a sense of fragmentation.

The number of iteration steps is between 28-35, and CFG defaults to 7.

You can open AD for facial repair.

The adaptation of 2D and 2.5D models is better, other models can be tested by yourself!

Welcome to return the picture and give me more comments on modifications.