Convolution crashing horribly
Closed this issue · 4 comments
The biggest current issue that I have with "estimate_noise" (the wrapper) is that the convolution crashes horribly (segfault). I'll dig into this next.
The last few commits try to remove the MAD/STD dichotomy (just reduce it to one line inside the same program) and modularize things like smoothing. I think the setup makes sense (though I see that the method string might be unappealing) and I think we're close. I sort of corralled and abandoned the general distribution part for now.
Let me revise - the convolution with the beam goes great.
Bizarrely, the 2d median filter seems to create disasters. I'm getting malloc errors and then segmentation faults. I've tried a few things but no solution yet. There are other options to median filter, though I do not think that they are as fast.
No, it's not that either. Madness!
Tentatively fixed... seems to have been in the plate scale calculation. Cross your fingers.
I'm now more confident that the new plate scale fixes things. closing.