I think it really is for light gathering than corner sharpness. That has always been the intention of the micro lenses was to gather light. But has nothing to do with actually sharpness of lenses
As the development of electronic products such as digital cameras and scanners progresses, the demand for image sensors increases accordingly. In general, image sensors in common usage nowadays are divided into two main categories: charge coupled device (CCD) sensors and CMOS image sensors (CIS). The use of micro-lenses significantly improves the photosensitivity of the image sensors by collecting light from a large light collecting area and focusing it onto a small photosensitive area such as a photodiode
http://www.freepatentsonline.com/7473522.html
Basically micro lenses allow you to capture the light that otherwise would have fallen inbetween the photodiodes which are actually light-sensitive. Depending on the sensor the part of any given pixel's worth of area on the chip which was sensitive to light may only be 70% of the linear size (or half the area). By adding a micro lens the input size of which is the entire size of the "pixel" and funneling that into the the half of the area which is sensitive to light you maximize the number of photons you collect which means that for any given numerical ISO you have better signal-to-noise than you would have for the same sensor made without micro lenses.
To summarize the technical blah blah the micro lenses on the P30+ mean that at ISO 100 you have the image quality as the P45+ at ISO50 (though the later has more resolution because of the larger chip).
Even those getting down to the technical level of what micron size the "effective" pixels are can miss the fact that not the entire "effective" pixel gathers light. The jump from the P25+ to P45+ (both near-full-frame sensors at 22 and 39 megapixels respectively) was able to be made, in part because while the "effective" area of the pixel was reduced from 9 to 6.8 micron the are of the pixel which gathered light was kept roughly the same by shrinking (and modernizing) the electronics package which sits next to the pixel (and which is not sensitive to light). Throw in a superior path-to-Raw (the AD convertor, physical wiring, heat sinking, and black frame technology) and the dynamic range was actually increased (modestly but measurably) despite the decrease in "effective" pixel size.
For this and many other reasons not all pixels are created equal. I really wish a lot of this was more strait forward and could be easily and quickly explained without significant background knowledge. DxO tests have two problems; 1) they try to reduce the very multi-faceted element of IQ/noise/ISO to a single number and 2) they ignore the very large number of elements outside of the sensor.
Guys like Bob and I spend most of every day with this technical stuff and neither of us will claim to know half of what we need to to really, truly, fully understand why MFD handled correctly consistently outperforms dSLRs handled correctly.
That's why the company I work for, Capture Integration, spends so much time and money getting gear into people's hands for them to actually shoot. As with when we brought a full suite of gear for the Moab workshop. You really have to shoot it yourself (aided by advice from someone with lots of MFD experience) to see why people regularly drop $10-50k on gear which may share numerical attributes with gear which is $2-9k. The 5DII and P25+ may share the same resolution number, and the 5DII may be the best camera ever produced at that price point, but they are not even remotely in the same league.
*sigh* goes back to answering tech calls
Doug Peterson, Head of Technical Services
Capture Integration, Phase One & Canon Dealer |
Personal Portfolio