Hello, I test down the style conversion, the effect is not particularly good, there are some characters deformation, or some styles can not be migrated to the past, such as gipsy style, etc., why is this?

CrystalNeuro / visual-concept-translator

Code of ICCV 2023 paper titled General Image-to-Image Translation with One-Shot Image Guidance

Apache License 2.0

155 stars 12 forks source link

Hello, I test down the style conversion, the effect is not particularly good, there are some characters deformation, or some styles can not be migrated to the past, such as gipsy style, etc., why is this? #10

Closed xuboot closed 1 year ago

xuboot commented 1 year ago

Given a picture, I am sd1.5 generated or other model generated character picture

CrystalNeuro commented 1 year ago

The default parameters do not fit every style type. Therefore, you should turn the parameters described in README to achieve better results. Character deformation can also be mitigated by turning the attention injection ratio. Changing the stable diffusion base model may also work.

dmmSJTU commented 9 months ago

@CrystalNeuro @xuboot 你好，需要自己写测试代码吗？