Hi, thanks for the great work! Really appreciate the clarity and effectiveness of your method.
I have a question about the motivation behind your design choice — you directly treat π₀ and π₁ as the source and target image domains, rather than following the more common conditional denoising setup, where π₀ and π₁ would represent the noise and target domains, and the source image serves as the condition.
Could you share your insight on why this formulation works better or what motivated this design choice?
Thanks again for sharing such an interesting work!
Hi, thanks for the great work! Really appreciate the clarity and effectiveness of your method.
I have a question about the motivation behind your design choice — you directly treat π₀ and π₁ as the source and target image domains, rather than following the more common conditional denoising setup, where π₀ and π₁ would represent the noise and target domains, and the source image serves as the condition.
Could you share your insight on why this formulation works better or what motivated this design choice?
Thanks again for sharing such an interesting work!