View Single Post
Old 04-03-2012, 06:09 PM
philiphart's Avatar
philiphart (Phil Hart)
Registered User

philiphart is offline
Join Date: Jan 2007
Location: Melbourne
Posts: 1,060
i'm going to be a little controversial and say that i think the hype about improved high ISO performance with each new DSLR released (whether Canon/Nikon) is a little misplaced.

most of the gains are in noise smoothing algorithms applied to the JPGs only and which can be applied to files from any camera in post-processing anyway. there may be incremental (very small) gains in reduction of read noise and with the 5DIII a claimed increase in photoelectric conversion rate (which I'm betting is small) but i really don't think it is at all possible to increase the physical noise performance by two stops.

as i understand it, shot noise totally dominates most low light exposures, until you're shooting quite long exposures and thermal noise becomes significant.

i tested a 10MP 5D original against my 21MP 5DII and aside from a few more hot pixels which are easily handled (and not really *noise* in the true sense of the word), the noise performance of the original looks better to me.. not all that surprising given the much larger pixel size and therefore lower shot noise (gapless microlenses on the 5DII making the difference much smaller than the pure arithmetic of the pixel count would suggest). but the gapless microlens card has already been played now so little to gain on that front.

i expect the 5DIII will be *very* similar in RAW performance to the 5DII, as pixel size is almost identical. the noise i see in my images is not originating in the camera.. it is purely a statistical function of sampling low light levels with a given pixel size.

am i missing something.. do people really think there is scope to improve the noise performance significantly.. what source of noise is being reduced or how is the signal collection improved??

Reply With Quote