The 1.0 model is very powerful. If you are using it for the first time and have enough memory, it is recommended to download the 1.0 version!

Currently, there are many high-quality anime models available. Considering that it has been a long time since 2D models were last updated, and the quality of the original training set is uneven and generally blurry, we have reselected the training set. Some salvageable training images have undergone AI image enhancement, resulting in significant improvements in the upgraded basic model in terms of blurring in the distance, color, expression, and environmental generalization.

The recommended upscaling algorithm is R-ESRGAN 4x+

Recommendations for VAE: animevae (characters), vae-ft-mse-840000 (scenes)

CFG ratio: 3-7 (it is recommended to use 3CFG when creating creative essays, which may produce unexpected results. The higher the configuration ratio, the more realistic it may be, depending on the prompts, samplers, and steps)

Feel free to try other settings on your own, and we welcome you to share your work!