Hi Thorsten,
There is no bug here, although this topic is recurrent. The file you have uploaded is a 32-bit *signed* integer FITS image. It stores a 16-bit raw image as signed integer numbers in an extremely inefficient way: only 37969 discrete values have been stored in the immensity of the 32-bit range. You can verify this by selecting the image and entering the following command on PixInsight's console:
j Math.pow2( 32 )*($max - $min)
which will show the number of discrete values stored in the image. Unfortunately, there are some applications out there that insist in storing raw CCD data in 32-bit formats, which is both a conceptual error and an efficiency crime. Let's assume that the authors of these applications do this for marketing reasons, i.e. to give the user a false impression of extreme accuracy. Raw CCD data are 16-bit unsigned integer numbers and as such they should be provided by any decent acquisition software.
That said, after loading this image in PixInsight you have two options:
- Apply the Rescale process (with default parameters; just select the image and click the Rescale tool button). This will redistribute the data to occupy the whole numerical range available, but the image will lose any physical significance.
- Apply the following PixelMath expression:
($T - 0.5)*(2^32)/32768
with the Rescale result option disabled. This will restore the image to its nominal range, preserving its physical meaning (that is, respecting the original black point).
