Closed dprov closed 2 years ago
I might have missed which one I implemented, but your observation looks correct. I remember changing between two cases has really small effect on the smoothness of the depth, or at least the difference is very difficult to see.
Thanks for your reply, sorry I did not see it earlier. Indeed, it makes next to no differences, it was mostly about understanding what's happening.
In the VFS paper (and in the paper it cites) :
In contrast, in e.g. https://openaccess.thecvf.com/content_iccv_2013/papers/Ferstl_Image_Guided_Depth_2013_ICCV_paper.pdf and https://link.springer.com/chapter/10.1007/978-3-319-46475-6_36 , it's rather
However, in VFS's TGV solver, the way p is updated seems to follow these other papers :
Whereas I would have expected something like
Would you be able to comment on that? Am I missing something? Thanks