Open mshafiei opened 3 weeks ago
Sounds like something is broken, but before investigating I wanted to point out that the gradient w.r.t. the input image is going to be zero for that algorithm anyway, because it starts with a nearest-neighbor-sampled lookup table.
I'm actually interested in the derivative w.r.t. alpha. I think that doesn't rely on the nearest neighbor look up. Is that correct? I also tried gradient on other generators e.g. blur, bilateral grid but none of them seem to work.
Hi,
I'm trying to get the gradient of local laplacian generator. When I add
GRADIENT_DESCENT_OPT "GRADIENT_DESCENT"
topython_bindings/apps/CMakeLists.txt:48
the local laplacian app breaks. The output image turns out to be pitch black. Also the gradient image is black (you need to add an additional buffer as the last parameter list of the generator in thelocal_laplacian_app.py
). Is there any workaround for this issue?