masked = np.where(mask != 0) mask[masked] = mask[masked]/mask[masked]
Tuesday, April 3, 2012
Creating masks with GIMP
Last time I had to make some really fine masking, and I resorted to GIMP. At least under Linux it can read and save fits files, albeit it treats them as _images_, not _data_, which I consider fundamentally wrong in many cases. So, you open the fits file in GIMP, draw something white or black, using any tool and save it. However, there are many pixels that are not 0 or 1, meaning that the image cannot be used as mask. Here's a script to normalise it back. It is cautious, meaning all pixels that are not equal to 0 will be treated as ones (and your mask may get bigger than you were expecting, at least theoretically).