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

Question about method comparison #6

Closed
xk-huang opened this issue Sep 25, 2021 · 3 comments
Closed

Question about method comparison #6

xk-huang opened this issue Sep 25, 2021 · 3 comments

Comments

@xk-huang
Copy link

xk-huang commented Sep 25, 2021

Thanks for sharing your exciting work about portrait editing!
I have a question about the comparison between PIRender and StyleRig. As there is no open-source implementation of StyleRig, I'm wondering did you implement it by yourself or asked the authors to conduct the experiment? I'm wondering whether the implementation gaps affect the final outcomes or not.
Thanks again for sharing your work!

@RenYurui
Copy link
Owner

Hi,
We re-implement StyleRig on our own.
We use the implementation details provided by the paper. Some details are different. For example, StyleRig uses a different expression model which is not publicly available.
We find the trained model tends to generate images with a conservative strategy: motions away from the distribution center are weakened or ignored. We tried many times but fail to improve the results.
However, I think this may not affect the final conclusion (the advantages and disadvantages of the methods).
Yurui

@xk-huang
Copy link
Author

xk-huang commented Oct 3, 2021

Thanks for your kind reply!

@xk-huang xk-huang closed this as completed Oct 3, 2021
@lastapple
Copy link

lastapple commented Oct 3, 2021

对比实验我也很困扰,不仅stylerig不开源,PIE也没开源,PhotoApp更无,不是不能自己复现而是怕不太公平或者效果不一样。向作者发邮件也没收到回复,问了另外一篇有stylerig结果paper的作者也都是自己复现的。

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

3 participants