[RICHARD L.]

If you want to visualize what "DIFFRACTION" does to an image, below are two pictures, the left one taken @ f/16 and the right one @ f/45 on a Pentax 645Z + P67 90-180 mm f/5.6 zoom at 90 mm FL, on tripod.

Diffraction robs your picture of most of its "bite", making the whole image fuzzy and unsharp.

[Weevil]

f/22 which is past the diffraction limit

Quick question: How do we determine what is that "Diffraction limit" for a specific lens ?

[GUB]

Quick question: How do we determine what is that "Diffraction limit" for a specific lens ?

There tends to be an opinion of diffraction limit for formats rather than individual lenses.
F5.6 is about it for the Q, f8 for apsc and f11 for full frame. Or thereabouts.
In practical use all that means is that is the peak of lens performance for that format - it still may make sense to stop down a bit more . Like everything in photography a trade off - you trade off some in-focus sharpness to achieve more depth of focus.
But if you look at the mtf curve Clackers supplied in #16 you can see the drop off. I am guessing it is a chart for an apsc camera.
EDIT and here is one for a full frame camera and a vintage lens that only stops to f16. I think you can see f11 is the peak.

[dms]

I believe the diffraction limit on resolution = constant * wavelength/f-stop.
It (diffraction limit resolution) is not a function of the lens or sensor--but rather the aperture, but indirectly with lens design and sensor specifics, because a lower resolution lens and/or sensor (larger pixel diameter) does not see the diffraction limit, untill it is no longer lens or sensor limited (e.g., f/22). Thus,

-- a high resolution lens takes the hit at a larger aperture (e.g., f/11) than a lower resolution lens (e.g., f/16).

-- And sensor that has a larger pixel density (e.g., KP vs K-x, or Pentax Q vs K-5) is likely to have the diffraction limit at a larger aperture (e.g., f/5.6 vs f/11).

[swanlefitte]

The diffraction limit is based partly on pixel size. The lens starts creating less a point and more of an airy disc. If the airy part is small enough to fall on a single pixel it doesn't matter. When the point falls across several pixels it matters.

The same lens on a 6mp apsc sensor will hit the limit at a smaller f stop than a 26mp apsc as the pixels get smaller.

[clackers]

It (diffraction limit resolution) is not a function of the lens or sensor

That's right, it's not whether its FF or APS-C, it's about the pixel size.

The K-1 and the K-30 have the same pixel sizes, they're cut from the same sensor wafer.

[GUB]

That's right, it's not whether its FF or APS-C, it's about the pixel size.

The K-1 and the K-30 have the same pixel sizes, they're cut from the same sensor wafer.

So the K-1 and the K30 would throw a matching mtf curve for a given lens? (With the K-30 at a lower LWPH)

[clackers]

So the K-1 and the K30 would throw a matching mtf curve for a given lens? (With the K-30 at a lower LWPH)

Yeah, except that the curve extends further to the right with the K-1, since the sensor's bigger.

[GUB]

Yeah, except that the curve extends further to the right with the K-1, since the sensor's bigger.

Now yar on drugs !

[GUB]

I know it is pushing it to compare these two ephotozine tests - one with a K-3 and one with a K-1 but you can't help coming up with the conclusion these are quite different results. I wonder how much is from the pixel pitch and how much is sample variation and how much is inaccuracy.

[dms]

The figures on left may not start at 0 on the y axis. Some test results do this, which means they don't scale similarly to those starting at zero.
In fact they most certainly don't start at zero, as poor does not have 1/5 the resolution of excellent. It should be more like 1/2 to 2/3.

[RICHARD L.]

[swanlefitte]

.I wonder how much is from the pixel pitch and how much is sample variation and how much is inaccuracy.

That is interesting in 2 ways. The smallest pixels do better at smaller aperture. The smallest sensor does worse at the edges which is not even the edge of the larger sensor.

I don't see pixel pitch explaining this

[GUB]

That is interesting in 2 ways. The smallest pixels do better at smaller aperture. The smallest sensor does worse at the edges which is not even the edge of the larger sensor.

I don't see pixel pitch explaining this

Been thinking about it and maybe we are seeing the lens resolution potential being not realised with the larger pixel. So therefore the K-1 doesn't peak and also the k-3 has a higher sharpness to fall from as they say with diffraction.

The figures on left may not start at 0 on the y axis. Some test results do this, which means they don't scale similarly to those starting at zero.
In fact they most certainly don't start at zero, as poor does not have 1/5 the resolution of excellent. It should be more like 1/2 to 2/3.

agreed - it is the top profile I was noticing.

[dms]

The top profile is exagerated by not scaling to zero. If they start it at 2/3 of full resolution the top profile will show differences that are three times as large as the ones that start at zero. And if 1/2 of full resolution they will be exagerated twofold.