Originally posted by: ElFenix
once we're talking about printing the size of the pixels has very little influence on noise. which is why pixel peeping is silly. the two largest factors influencing noise are size of the sensor and then the technological features of the sensor (i.e. what process it uses, gapless microlenses, amplifier noise, etc.). which is why crop sensors are now able to compete with the original 5D, with it's 3+ year old tech. the other silly thing about pixel peeping is that lower numbers of pixels will always appear to have less noise because they are less accurate.
there's also a difference in type of noise & how the human eye perceives it e.g..
"We were discussing the noise characteristics of the A900 and a point that Andy made, with which I agree, is that unlike some cameras the noise on the A900 is stochastic in nature. In other words, less grid-like and more random. This makes it much less objectionable to the human eye, and so when noise characteristics are compared numerically this perceptual nuance is not factored in, yet it can mean a considerable difference in the way that we perceive noise. In other words, noise with the A900 at higher ISOs (at least up to ISO 1600 when chroma noise starts to kick in), is visually less objectionable than it is on some other cameras which claim lower noise. More film like, and less digital in other words."
http://www.luminous-landscape....ameras/sony-a900.shtml
I've seen reviews over the years where a camera whose output looked better on screen than anothers looked worse when printed. I guess that part of the maker's optimisation also includes expected viewing medium & they may differ.
.