The diffraction limit is not a cliff

Somewhere in the middle of learning photography you get told that stopping down past f/11 makes images soft, and after that you either avoid small apertures forever or you ignore the advice completely. Both reactions come from the same misunderstanding: that there is a specific aperture where diffraction switches on.

There isn't. Diffraction is present in every photograph you have ever taken, at every aperture, and the question is never whether it is happening but whether it has become the thing limiting your result.

What is actually going on

Light passing through an aperture spreads. A point source imaged by a perfect lens does not land as a point but as a small bright disc surrounded by faint rings, the Airy pattern, and its size depends only on the wavelength of the light and the f-number.

The diameter of the central disc is 2.44 × λ × N. Taking green light at 550nm:

Two things are worth noticing. It scales linearly with f-number, so every stop down multiplies the blur by about 1.4. And it has nothing whatsoever to do with sensor size, focal length, or lens quality. An f/11 Airy disc is 14.8µm across on a phone, on a full frame body, and on a large format camera. The physics does not know what is behind the lens.

Why sensor size changes the answer anyway

If the blur is the same size everywhere, why does everyone say small sensors hit diffraction earlier?

Because what matters is the blur relative to what is recording it. Compare the Airy disc to the pixel pitch:

A 24 megapixel full frame body at f/8 has a 10.7µm blur spread over 6µm pixels. A 20 megapixel Micro Four Thirds body at the same f/8 has the identical 10.7µm blur over 3.3µm pixels, so it is smeared across three times as many of them. Same optics, different consequence.

The usual working definition is that a lens becomes diffraction limited when the Airy disc grows past roughly twice the pixel pitch. By that measure the 24MP full frame body is fine to about f/13, the 61MP body to about f/8, and the Micro Four Thirds body to about f/6. Those numbers are why the same advice sounds wrong to different people.

And why print size changes it again

Pixel pitch is the strictest standard available, and most photographs are not viewed that way.

If you use the traditional circle of confusion instead — 0.03mm, or 30µm, on full frame, the value that corresponds to an 8x10 print at normal viewing distance — then diffraction only becomes visible when the Airy disc exceeds it, which happens around f/22. That is the origin of the old advice that full frame is fine to f/16 and questionable at f/22. It was correct for prints. It has never been correct for 100% inspection of a 45 megapixel file, and the gap between those two standards produces most of the disagreement online.

This is the same output-dependence that makes depth of field slippery, and for the same reason. Both questions are really "is this blur visible", and visibility depends on how large you display the picture. Why your depth of field calculator disagrees with your photographs works through the other half of it.

The curve you are actually riding

Here is the part that the "f/11 cliff" framing gets backwards.

At wide apertures your lens is limited by aberrations — spherical, coma, astigmatism, field curvature — and these get better as you stop down, quickly. At small apertures the lens is limited by diffraction, which gets worse as you stop down, steadily. Real sharpness is the combination, so it rises from wide open, peaks, and then declines.

For most lenses that peak sits two to three stops down from maximum aperture. An f/1.4 lens is usually at its best around f/4 to f/5.6. An f/4 zoom is often best at f/5.6 to f/8. Past the peak you are giving up sharpness at the plane of focus, gradually, in exchange for more depth.

Gradually is the important word. Going from f/11 to f/16 does not ruin a photograph. It costs you a modest amount of contrast at fine detail. Whether that is a bad trade depends entirely on what you get for it.

When stopping down past the limit is the right call

Often, and this is where the internet advice does real damage.

Diffraction blur is uniform across the frame and mathematically well behaved, which means capture sharpening and deconvolution recover a meaningful portion of it. Defocus blur from insufficient depth of field is neither uniform nor recoverable. Given a choice between a frame that is slightly diffraction softened everywhere and a frame with a crisp middle distance and a mushy foreground, the first is easier to work with and usually looks better.

So: if f/16 is what it takes to hold the foreground and the horizon together in a single frame, shoot f/16. If f/22 is what it takes to get the whole of a small object in focus and stacking is not possible, shoot f/22. The diffraction calculator will tell you where your particular sensor crosses the threshold, and that is useful information, but crossing it is a decision rather than an error.

The cases where you should genuinely stop are narrow: when you have room to open up without losing anything you need, and when depth of field is not the constraint at all. Stopping down "for safety" on a subject that is entirely at one distance costs sharpness and buys nothing.

Macro, where it arrives much earlier than you expect

Close focus changes the sum, and it changes it a lot.

As a lens focuses closer it extends, and the effective aperture becomes N × (1 + m), where m is the magnification. At 1:1 the multiplier is 2, so a marked f/8 is behaving optically as f/16, with an Airy disc of 21.5µm rather than 10.7. At 2:1 the marked f/8 is an effective f/24.

This is why macro photographers hit diffraction at apertures that landscape photographers consider mild, and why the marked f-number stops being a useful guide down there. It is also, more than any other single factor, why focus stacking exists as a technique: at high magnification the depth of field available at a usable aperture is often a fraction of a millimetre, and stopping down far enough to fix that destroys the detail you came for. The magnification calculator reports effective aperture alongside reproduction ratio for this reason.

Practical settings, with the reasoning attached

On full frame, landscapes at f/8 to f/11 sit close to the sharpness peak of most lenses while giving substantial depth. Go to f/16 when the composition genuinely needs it. Treat f/22 as a tool for specific problems — long exposures without a filter, sunstars, extreme near-far compositions — rather than as a general purpose setting.

On APS-C and Micro Four Thirds, shift roughly one and two stops respectively in the wider direction, both because diffraction bites earlier and because those formats already have more depth of field at the same framing.

In macro, work between f/5.6 and f/11 marked, watch the effective aperture rather than the marked one, and stack rather than stop down when the subject allows it.

And with any of it, the useful check is your own. Photograph a detailed static subject at every aperture on a tripod, compare at the size you actually print or publish, and you will find your lens's peak and your own tolerance in twenty minutes. That result beats any general rule, including this one.