Quote:
Originally Posted by Merlin66
This discussion came from some discussions where separate 8 bit (camera) images were stacked (summed) - the claim was made that the large ADU count in the sum required greater image bit size usage.
I then said I use 10 bit SER Video frames and stack hundreds of them (solar imaging)- does this mean I get the equivalent of a 98bit image??????
(I know when they are stacked (median combine) the result is still a 10 bit file)
The answer given,.....yes ofcourse!!
He would not be convinces otherwise.......
|
Actually - if you stack, say, 1000 frames you only need another 10bits (2^10=1024) or 20 bits in total to store the sum. If you retain 10bit precision by averaging then you'll be adding in some noise from rounding.
It gets more interesting above 32bits when floating point becomes a viable option. With 32bit floating point The IEEE 754 standard provides 24bits precision and exponents from -126 to +127. Plenty good enough for AP. Or you could stick with 32bit integer and sum up to 65535 16 bit images with
no loss of precision.