megapixel vs. lens resolution?

Started Feb 1, 2010 | Discussions thread
GordonBGood Veteran Member • Posts: 6,308
It's all a matter of the photosite density...

Comments as follows:

Doug Pardee wrote:

hotdog321 wrote:

Would a FF 35mm camera with 50 megapixels be the ultimate? 100 megapixels? At what point does it become meaningless because even the best lens cannot resolve any finer detail?

There is no such point. Both the lens and the sensor contribute some amount of blurring. Making either one sharper makes for a sharper image.

It certainly is possible to reach a point where the difference that a sharper sensor makes is so small as to be irrelevant (a drop in the ocean). We're pretty much already there in terms of normal-sized prints. A 12-megapixel 4:3 digicam produces 4000x3000 pixels. Cropped onto a 5x7, that's about 570 pixels per inch, which is way more than the ~300 pixels per inch resolution of most printers. But cropped onto an 8x10 it's 375 pixels per inch, so the camera is still contributing maybe half of the sharpness while the printer is contributing the other half.

The question is, isn't there such a thing as sharp enough? Using sharpness as being the primary technical measure of a photo doesn't seem reasonable to me. For that matter, using technical measures as the primary consideration for evaluating a photo doesn't seem reasonable to me.

Skipping to the chase, if we consider that even an ideal lens will be limited by the physics of diffraction at its widest aperture and that according to the Diffraction Airy Disk calculator at , an f/2.8 aperture has an airy disk of 3.8 microns scaled linearly with aperture value (ie. f/1.4 has an airy disk of 1.9 microns), and also that Emil Martinec (ejmartin of these forums) has shown that one can increase the pixel spacing up to about twice the photosite spacing as the diameter of the airy disk before there is no more effective gain in detail, we can calculate the photosite density so as to be limited by the maximum aperture of a camera's lens. For the f/2 lenses of the Panasonic LX3 and Canon S90, the limit would be about 55 MP per square cm. where as these are at densities of 24 and 23, respectively, meaning that they could be increased to about 24 MP to have them definitely limited by the lens diffraction as far as any gains in image detail. However, although that would eliminate moiré patterns for the green photosites that largely influence luminance patterns, there would still be the possibility of colour moiré patterns especially for the shorter wavelength blue photosites due to the red and blue being sampled at half the horizontal and vertical frequency for a normal Bayer pattern sensor; thus one might have to quadruple the pixel density in order to completely eliminate the effect.

The latest 1/2.3" sensors with their 14 MP are very close to that limit already at about 50 MP per square cm. and especially considering that many of these cameras have a f/3.3 maximum aperture are well past their theoretical resolving power of about 20 MP per square cm. These cameras do not need an Anti Aliasing (AA) filter and if the MP ever increases to about 23 MP for a 1/2.3" sensor there would be no possibility of any moiré patterns using an aperture of f/3.3.

Even if we don't consider wider lenses as they are likely limited by other things other than diffraction, this same photosite density allso applies to the larger sized sensors so that a full frame 35 mm. sensor would need to have over 400 MP in order to have this same diffraction limit at about f/2.8 or so (not including the extra colour moiré patterns), where likely lens distortions would be the limit even before this point.

Regards, GordonBGood

Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow