Nokia's white paper adds, "What’s more, based on Nyqvist theorem, you actually need oversampling for good performance." [/QUOTE]
The Nyquist criterion requires that you filter your signal so that you'e not aliasing higher frequency information down to lower frequencies. What they're doing is removing the antialiasing filter by increasing the spatial sampling frequency. I still say that for the lenses they're using all they're doing is getting a higher resolution of the blur disk - no AA filter required for that!
nice one, thanks Nevro.
(I think we better stop)
I wouldn't be so sure. Mobile phone optics providers publish MTF charts at an incredible 400 lp/mm (over 10 times higher resolution than the highest that Canon publishes) -- and even the cheap ones still have usable contrast at that spatial frequency. (I don't have the link right now, but one of them was posted to Image Sensors World a few years ago, no doubt optics have improved a little since then.)
I tend not to consider a camera phone decent unless they've gone to some measure to protect the lens. Sony Ericsson were good for that a few years ago.
Zeiss F2.0 lens - interesting. I am looking forward understanding this a bit more.
If you see me with a wrench, call 911
I still think a scientific study is in order, throwing many different brands of phones out of the window driving at different speeds, to eliminate any statistical errors.
I'll volunteer to be the driver, everyone send me your iphones and i'll happily send them flying at 200km/h :P
An awful lot of electrons were terribly inconvenienced in the making of this post.
Gear Photos
A decent read describing the Nokia's sensor and their oversampling / digital zoom:
http://arstechnica.com/gadgets/news/...one-camera.ars