gnthibault

Ok, cool I think I managed to do what I was looking for, ie, fit a bivariate gaussian on reconstructed psf. However, I now need to understand this statement: `...

Now I understand the limitations of my current approach: -It is not clear how the bouding box for start patch selection is chosen. On my test, the patch size was...

I am so thankful ! thank you very much ! I have also started a notebook on this very topic (basic level though): https://github.com/gnthibault/Optimisation-Python/blob/master/SpatiallyVariantDeconvolution.ipynb For the moment, I am using...

Ok, looks like a good approach would be to compute the "goodness of fit" either based solely on the residual or using a statistical test (bayes factor ? something more...

Ok from what I saw from tutorial 3, I think I already managed to find the coefficient map and reconstruct the psf at arbitrary location following your tutorial or looking...

Ok here is what I got checking few examples of source detection with tutorial 3 example. That correlates with what I was seeing in terms of reconstructed psf:  There...

Ok I will document my trial. In the meantime, can you check the _bkg versus __bkg atteibute in single_image class ? In my understanding there might be a bug there.

Ok I did not realized it was on purpose, I will check again then because at some point I saw that source detection was maybe not using background substraction beforehand....

Hi @BrunoSanchez and thank you for your feedback. I used this very simple RGB image (non calibrated filter) to check if the computed psf ellipticity was following a radial pattern....

from skimage import io array = io.imread('path/to/data.jpg')