Quote:
Originally Posted by jjjnettie
I beg to differ on that point.
Refer to the 2 images I've posted.
They are the autosaved files from DSS, converted to 16bit, cropped, resized and saved for the web.
The only difference between them is that one has had darks taken (not ICNR) then subtracted, the other, no darks at all.
Take a look and then tell me if you see much of a difference. 
cheers
jjj |
Dear JJJ
I'm not sure what you mean.
If you take an image with a DSLR and save it as a jpeg, a compression algorithm is used to store the image. Depending on the amount of compression this will smooth out areas of the image with pixels with similar counts-similar but not the same.
If you take a dark frame and save it as a jpeg then the same process occurs. The problem is that the compression algorithm will not necessarily smooth out the data the same way in both images. This means that if you convert both images to uncompressed files and subtract one from the other then there will be some error as induvidual pixels are not necessarily being subtracted from the same pixels.
A raw (or fits) image file is literally a table of all the counts detected in each pixel. A dark file is the same thing. As a CCD is being exposed as a dark, electrons will accumulate in the induvidual pixels at varying rates depending on the temperature and the individual characteristics of each pixel. This is the dark current and should be fairly constant for each pixel. This is why they can be subtracted from the light images to remove the variation in background. This needs to be on an individual pixel basis to be accurate.
I assume the ICNR method does this before the image is converted to a jpeg. As RAWs you can do it afterwards. The advantage to this is that you can average (or median) many darks to get rid if random noise etc and it wastes less time aiming at your subject.