Masking in deep can be a whole other can of worms, yes.
Basically every time you are dealing with transparency of some sort
things get a little more complicated as you need to be able to define
how each sample that contributes to the transparent pixel has to be
treated to deliver predictable
Hello!
I am running into some issues with the DeepTransform node and was wondering
if anyone has experienced similar behavior.
Basically the situation is that we have some shots that are very nearly
lockoff shots, just with some subtle nodal camera movement. In order to
save on disk space we are
From what I remember, DeepTransform does not filter like you would
expect from a 2D tranform, as that would be a potentially crazy amount
of calculations (every single sample would have to be filtered and there
could be lots).
Hence this node should be used with care for things like