Angular resolution of a lens and pixel size. The formula of 206*pixel size in microns. Divide answer by focal length in question. That will give you the coverage each pixel has.
As for raw angular resolution, Dawes Law. Example, a "perfect" 160mm diameter objective (larger than the ones found in the Canon 600/4 but smaller than 1200/5.6L EF) will give about a .74 arc second resolution. A 101mm (4 inch) objective will give about a 1.1ish arc second resolution. These are "ideals" and very few lenses will reach the resolution. Which brings us to the third area.
Airy disc size. If you were to take a f/1.0 lens, the airy disc (for a PERFECT ONE) will be 1.0 * 1.44, which would equal a 1.44 micron sized disc. That's for a f/1.0 lens with a PERFECT strehl ratio. Commercially made lenses DO NOT have anywhere close to a perfect strehl ratio. Basically it's how much energy is concentrated into the airy disc vs wasted energy into secondary and tertiary rings. You lose contrast, sharpness,etc with a commercially made lens. Strehl ratios are less than 0.8, with a 1.0 being a perfect optic. Scientific optics, like those made at Astro Physics, Takahashi and Telescope Engineering Company are in the realm of .95 or better. The TEC and AP are usually .98x or better. As for Canon, they don't list a strehl ratio or people will be comparing numbers to one another. But the strehl is below 0.8, or that of the diffraction limit. Leica supertelephotos were the only lenses that I know that had a diffraction limited rating (0.8) in their literature.
So, if you take a perfect fictional 300mm 1.0 (just for fun!) with a sensor, you would need 1.44 micron sized pixels to represent the airy disc correctly. And it would have an angular resolution of better than .5 arc seconds. The limitations right now are the pixel sizes for the cameras as well as the strehl ratios and angular resolving capability of each lens. A 300 f/2.8L EF will outresolve a 300 f/4 or 300 f/5.6 due to Dawes Law if they all had equally perfect optical formulas. 100mm aperture vs 75mm vs 53mm. Larger the aperture, more angular resolution.
We haven't come close to resolving the full power of the lenses. Make a sensor platform with 1.44 micron sized pixels. Then compare them with say a 5.7 micron sized pixel.
Now here's the rub. If you're always shooting at f/5.6, you won't see the difference. At f/5.6 you're looking at, on a perfect lens, an airy disc of about 8.1 microns. Now try the lens at f/2.8, you'll need a 4 micron sized pixel to sample a perfect f/2.8 airy disc. The 7D comes very close though at 4.x microns. But if you're taking about a f/1.0 lens, we still have a ways to go.
In other words, it also depends on the f/ratio that you're shooting.
SO, F/ratio (airy disc), lens size aperture (diameter for Dawes Law) and how well figured the lens is (strehl ratio) all come into play.
Hope this helps.
woos wrote in post #14607414
Sort of, but the max angular resolution thing isn't really something one needs to worry too much about in the photography realm. That's something that is going to be more pertinent with telescopes, really, imho. You can try to find a point where the lens no longer resolves any contrast difference between lines super close together and try to extrapolate "how many megapixels the lens resolves" from that, but it's a stretch. What color are the lines (this is one of the huge gotchas that makes the diffraction stuff not apply as much in the real world as people think it will in most cases)? What do you consider to be an acceptable amount of contrast to count as still resolving the detail? What aperture and what part of the image?
Also, say you have a lens that resolves less detail than the sensor can sample...until you reach an absurdly low level of detail from the lens, something like HALF the potential resolution of what the sensor can capture, having more megapixels still results in more detail being captured
. It's probably even worse than that, taking into account the bayer array (the other reason diffraction isn't as visible in the real world on say, the d800, as you would think by plugging the number into one of those dozens of online diffraction calcuators---there are a couple that mention this stuff though)...
Really we aren't anywhere near the MP of the camera not delivering more resolution. 36mp isn't enough to get all the detail possible from a decent modern lens. Yeah, there's going to be some soft far corners and such out there on wide angle glass, but for the most part it's not a big deal. We'll be up to at least 100mp before we're really getting all we can from our 35mm size sensors. More samples also helps with lens corrections, I suspect Canon's DLO is forward thinking in this respect.