Originally posted by L33tGreg Right, 2 times more possible integer values. but going from 14/bpc to 16/bpc is an increase by 2/14 or 1/7 or about 14%. That's the increase in the file size. However, my guess is that most scenes wouldn't have this much DR so with compression, the extra bits would add negligible size and value. But on scenes with that much DR, you would see at most a 14% increase in file size.
*I am a computer scientist...
Greg
Hi Greg,
i think the problem would be noise, as it isn't really compressible, unlike actual image information. (unless somebody likes to take photos of random noise, but most image motives look rather well organised) So if you have a raw file of a sensor which isn't really capable of 16 bit, then you would have a nice, compressible image in the top 14bits, while there ends up a huge amount of noise in the bottom 2bits, which would throw off your 2/14 figure for the added file size. (depending on the quotient of compression rate for image information vs. noise)
(since my approach is theoretical as well, it would be interesting to know how much the difference between a 12 and 14 bit raw is, but i'm not sure how to work that out)
Alex