Hi,
I think it is more complex than just a few factors. The obvious factor are:
- Noise as you said a bigger sensor surface collects more photons, leading to less shot noise
- MTF, any feature on a MF sensor will be larger assuming same FOV (Field of View) so MTF will be higher
Now, the less obvious factors
Are MF lenses better? Maybe, or may be not. It is probably easier to build a good 135 lens than to build an MF lens. With SLR designs the mirror assembly requires telecentric designs on short focal lengths. More symmetrical designs may have better corner performance, but we get problems with excessive lens cast.
DSLRs are usually used with zooms and MF more often with prime lenses.
What I would expect?
Well, it seems there is a market for low end MF digital, like 22-25 MP backs. I'm not sure people are buying them for image quality.
For the midfield, Pentax 645D and the Leica-S2 and others I would presume that the Nikon D800/D800E is a real competition. I guess that its is practical people buying the 645D and it is a fine camera. It lacks Live View for critical work. It's a hard to ignore fact that you can buy two D800/D800E bodies and a bunch of very fine lenses from Nikon and Zeiss just for the price of the 645D body.
Leica S2 is a different thing. It has a red dot and awesome lenses, but it still lacks live view.
Than we have a high end like Phase One IQ180. These make a lot of sense if you have the money and the knowledge how to drive them.
In my view the Nikon is D800 is not the ultimate camera. We already have 24 MP sensors in APS-C, and they are just doing fine. I'm pretty sure we will see 50MP full frames in a couple of years). Perhaps a Nikon D4X. Will the lenses keep up? Good question. The best primes are good for 50MP FX but most zooms will have bad corners.
Finally, I think that MF digital needs live view badly.
Best regards
Erik
Maybe but not necessarily. For one thing, the gap in sensor performance might not grow much more: the best CMOS sensors now seem to be fairly close to theoretical limits on maximum sensitivty (for a CFA sensor) and low dark noise. Also, with full 645 format of 56x42mm (which is larger than 36x24 by about 75% linear, compared to only a 28% size advantage for 44x33), the inherent advantages in the quality of the images from a larger sensor (like resolution about 75% greater from the same basic lens design scaled up) might mean that good 645 lenses with CCD sensors keep an advantage over 35mm format lenses even with a perfect sensor.
In fact, I am sure that some people believe that even 44x33 and 45x30 is big enough that the inherent advantages of the lenses keep those formats ahead in IQ, but the recent comparisons between Nikon D800 and Leica S2 suggest that the 25% size gap is not enough.
EDIT: another point in favor of larger formats that I overlooked. Even if dynamic range is better for a smaler CMOS sensor, that only relates to how much noise is seen in deep shadows (and with severe underwxposure ofmtue sensor, meaning at higher ISO settings.) When instead you look at fairly well lit parts, maybe all the normal printable range from the highlights down to three or four stops below midtones, the dark noise used in DR measurements is overwhelmed by photon shot noise, and looking at just that shot noise, a larger sensor that counts more total photons has an advantage in signal-to-noise ratio. This perhaps relates to the finer, smoother tonal gradations often claimed for MF in comparison to 35mm format.