Gaussian blur question

Yes, of course.
Actually, this is really something you could guess by studying the source code :slight_smile:

You are right . The reason I have so many doubts is due to cimg macros that are scattered throughout.

Thanks again for answering all my doubts .

Hi @David_Tschumperle more inpaint answers please :slight_smile:

  1. When looking for offset centers to search for :
    a) Do we only look at those reconstructed centers that share pixels with the target patch ?
    b) Do we look all around the offset center when performing lookup increment ? From the code it suggests that while searching for more patches we only increase x and y by lookup increment and not look backward (decrementing lookup increment). That sounds like we would miss some source patches ?

Thanks!

Ping @David_Tschumperle

I wonder what all the inpaint stuff has in common with the ‘gaussian blur’!

1 Like

Nothing. Just happened to ask on thr same thread