Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

compared to PASD #47

Open
Feynman1999 opened this issue May 17, 2024 · 1 comment
Open

compared to PASD #47

Feynman1999 opened this issue May 17, 2024 · 1 comment

Comments

@Feynman1999
Copy link

为了方便,直接用中文了~
seesr相比PASD的改进是不是主要是在representation branch上(针对低质量图片做了训练),sd和controlnet部分差不多?
另外就是seesr没有像PASD那样在controlnet的输入上做显式地增强?

@cswry
Copy link
Owner

cswry commented May 26, 2024

SeeSR相对第一版PASD的主要改进:
(1)使用tag风格的text prompt,能高效地激发SD先验对local object的生成能力。
(2)degradation-aware text prompt提取器。
(3)将LR latent引入diffusion起点来提升结果稳定性。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants