During stacking, Siril seems to cut negative values, in order to keep the range of the data between 0-1 or 0-65536 which is normally is not a problem because you don’t have negative values in your images.
But when calibrating images (especially darks), you have to subtract the offset from it which might result in negative values because both are very dim images. This should be normal but it seems for some cameras (Canaon EOS 700D and 6D tested so far), after subtracting the offset/bias from the darks, most of the values are actually negative:
Even the median is negative which means that the majority of the values in the image is negative, so if the negative values are cut, most of the information is gone.
I would say that all the values < 0 are cut because in the result there are no negative values and maximum hasn’t changed:
which would happen when the neg. values are for example shifted over 0 by adding in this case like 485 to all values. If I subtract the offset from the stacked, not calibrated dark e.g. in the PixelMath tool or stack the calibrated dark in Fitswork for example, the values are not cut:
Siril seems to handle images with values < 0 during calibration just fine, when I use the dark with neg values for lights and flats, the resulting image looks less noisy then the image with dark where neg. values are cut during stacking:
full range:
with negative values cut:
The difference is not super big, but clearly noticeable. In the lower image the background is noisier and you can see small and short strokes/lines in the background.
My question is now if this behavior is intended and if so why it is intended?
cs,
Pip