Ulrik, the other manufacturers don't have enough invested in these lenses! Introducing a new system while there is a back catalog for the competitor, without discontinuing their SLR system is a bad idea.
We might see it, but only with a new mount or perhaps emulating the M mount, but we won't see an S-mount.
I managed to rescue my d700 files, but realize that the leica lens is... somewhat better than my old 24 ai-s.. :P But i guess you guys get somewhat of a picture of what the camera is capable of, going to go down and test the sigmalux vs. a summilux tomorrow, instead. Since I think most of the softness in the d700-file is due to lens and focus. But you get some kind of impression of amount and nature of noise? Anyway, sorry for this halfass comparison, I'll make a better one tomorrow :P
Just contemplating... how valid is such a comparision if there's in-camera NR going on? Smearing or lack of sharpness at higher ISO's seems to suggest that that's what Nikon is doing regardless of how it's denied. Your Nikon lens may well be soft, but I suspect it's sharper at low ISO (Would be interesting to test that as well)
ulrikft wrote:
I managed to rescue my d700 files, but realize that the leica lens is... somewhat better than my old 24 ai-s.. :P But i guess you guys get somewhat of a picture of what the camera is capable of, going to go down and test the sigmalux vs. a summilux tomorrow, instead. Since I think most of the softness in the d700-file is due to lens and focus. But you get some kind of impression of amount and nature of noise? Anyway, sorry for this halfass comparison, I'll make a better one tomorrow :P
Just curious how you did this test. Are we looking at 100% native crops from both cameras or have you re-sized the M9 down or D700 up to match the other?
And I'd like to ask about the conversions... In camera jpg?
Actually the M9 shows VERY obvious signs of NR, or a very strange raw-conversion. Check out the blue and red channels... full of "lumps" and totally flat areas. This does NOT exist in nature, that has to be either in-camera noise reduction or the raw-conversion. The D700 on the other hand shows a very natural gaussian-distribution noise in all three channels, which implies that no NR has been applied.
theSuede wrote:
Thank you for the comparison, ulrikft... :-)
And I'd like to ask about the conversions... In camera jpg?
Actually the M9 shows VERY obvious signs of NR, or a very strange raw-conversion. Check out the blue and red channels... full of "lumps" and totally flat areas. This does NOT exist in nature, that has to be either in-camera noise reduction or the raw-conversion. The D700 on the other hand shows a very natural gaussian-distribution noise in all three channels, which implies that no NR has been applied.
I'm gonna go out on a limb, and maybe stun some people here, but CCD sensors aren't ideal for low light shooting.
CMOS sacrifices colour accuracy and dynamic range at low iso's to get lower noise at high iso. CMOS sensors also don't get as hot.
The reason for the difference in dynamic range is the size of the pixel (p-well) due to architectural considerations. Each photosite on a CMOS sensor needs space for it's individual amplifier, so the photosensitive surface area is less. Microlenses help counter this by directing light from a larger surface area into the individual pixel, but this increases luminance, not dynamic range.
CCD sensors on the other hand have row or column amplifiers, this is the reason we often see bright point lights on very cheap CCD cameras (especially video cameras) blow out the whole column as it overloads the amplifier for that column.
CMOS sensors are faster (on-chip processing), cheaper to produce, and also consume less power hence the epic battery life on cameras like the 5D Mark II.
So essentially the only benefit of CCD's is a bit of dynamic range and generally better colour and tonality.
Dynamic range in practice is the ratio of the highest signal level which can be captured to the level of the noise floor. The size of the photosite is not a direct factor. Yes, microlenses do not affect the DR of the photosite. I have to assume that CMOS photosites have lower capacities, higher noise or perhaps both.
Actually, colour accuracy is a legacy of the Colour Filter Array, not the sensor type. Most CMOS sensors end up with weaker CFA's to improve high ISO performance at the expense of colour accuracy as that's been where the market has been pushing things. The Sony A900 and Leica's are exceptions, which is why they both show such excellent colour accuracy and channel separation. MF Backs also have strong CFA's.
Size of photosite (in part) determines the amount of photons that can be captured before clipping, hence it is directly related to dynamic range. Do you have a nanotech background? Please explain how my statement is incorrect if it is. Here is an article explaining this phenomenon.
Colour "accuracy" is indeed a factor of CFA's. Effects like posterisation and pixel to pixel non-uniformity are better supressed on CCD's though, as each pixel on CMOS sensors have open loop output amplifiers, balanced against a "black frame". This is part of the reason CMOS has lower noise.
This is all very interesting banter but I think a little off topic.
Just rob a bank 
