Originally I wanted to create an extreme desire base model, but since the training is for a 2.5D model, the extreme desire effect cannot be achieved.

Fortunately, the model itself can directly generate 2.5D character images.

Due to the inferior quality of the 1.5D model compared to SDXL, the focus was originally on creating XL models.

However, there are too few XL users, it is still not mainstream up to now, there is no other way.

This model is quite good for generating 2.5D character images.

The 1.5D model still remains the main model for novel text recommendation.

So, this should still be popular.