All of which flies in the face of the observed fact that the best looking images seem to come from lenses which do not meet the resolution capabilities of the sensor. In other words, the sensor needs to exceed the lenses resolving power in order to get the best possible image from the lens. At just what point it becomes overkill is open to debate.
-- graywolf http://www.graywolfphoto.com http://webpages.charter.net/graywolf "Idiot Proof" <==> "Expert Proof" ----------------------------------- Powell Hargrave wrote: > Some interesting comments I lifted from the ProRental list. > > Powell > > ===================================================================== > > Date: Tue, 08 Aug 2006 02:36:26 -0400 > From: Steven Inglima <[EMAIL PROTECTED]> > Subject: [ProRental] Resolution and the limit > To: [EMAIL PROTECTED] > > In response to Lee: > > <All other things being equal - yes, larger pixel sites means more > volume of light captured and that means better signal to noise ratio and > greater dynamic range. Smaller pixel sites also affect the ability of a > given lens to resolve details to the resolution of a chip. > > All else being equal, of course, this is correct. What can change the > equation is the quantum efficiency of the sensor itself. If a given > sensor converts, let's say, 20% of the photons into electrons, and the > gathering area of the photo site is reduced by 20%, then should expect > to lose dynamic range as the shadows will be getting relatively little > information. But, if a different sensor can achieve a conversion of 40% > of the photons to electrons, then it's possible to shrink the photo site > and retain the same dynamic range. > > We are already at the limit of resolution with many of the lenses > currently available and more pixels crammed into the same size chip are > simply not going to offer much more that bigger file sizes! > > > There are rumors flying around about a 22mp Canon - even the very best > of the current Canon lenses would fail to perform at the resolution of > such a chip -- bigger file size but the same image detail and a good > chance that there would be lass dynamic range as well. Doesn't sound too > promising to me! More sophisticated noise reduction could help with the > dynamic range part of the equation but it would require much more > advanced optics to provide the needed extra resolution and I don't see > that happening any time soon, certainly with popular zooms! > > > Without confirming or denying the existence of an upcoming 20mp+ Canon, > the math of optical requirements is something that we can demonstrate. A > 5 micron photosite would require 100 lpm resolution from the lens. While > this is indeed a high benchmark, there are a number of lenses that Canon > makes today that can achieve this. And of course, if a slightly larger > micron photo site, let's say 5.5 would be employed, that would lower the > resolution requirement. So, in other words, you can't necessarily know > what's about to happen performance wise in digital photography on the > immediate horizon :) > > 1/ (2x .0055)= 90.9 lpm requirement. There are quite a few lenses in the > Canon system that can achieve this resolution. > > You can read about the Nyquist limit and other Fourier equations at : > > http://mathworld.wolfram.com/NyquistFrequency.html > > Or, you can just keep posted...cause Canon isn't likely to announce that > they will never make a camera with more resolution than the 1Ds mkII. > > Best wishes, > > Steven Inglima > -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net
