Why is my lens sharpness much softer at f/1.2 than higher apertures when taking landscape pictures?

Question

I recently bought a used Canon 50mm f/1.2 L USM and have been noticing that the lens is quite soft at f/1.2 (using a Canon 1DX Mark II, if relevant). Now it's the first time I am shooting at such apertures and am wondering whether what I am seeing is to be expected or if there is something off with the lens.

When focusing at the hyperfocal distance (auto-focus) to capture trees and buildings in the distance (>50m) I am seeing the following, noticeable difference between f/1.2 and f/2.8 (objects are in the centre of the image, zoomed in at 200%, RAW file):

f/1.2 with shutter speed of 1/8000 and ISO 100

f/2.8 with shutter speed of 1/2000 and ISO 100

Question: Is this difference in sharpness between the two apertures to be expected given the scenario?

Answer 1

Notice that the spruce tree in the first image has a lot of purple/red fringing to it.

That is an example of uncorrected longitudinal chromatic aberration (LoCA); where different wavelengths bend to different degrees due to their different frequencies. This picture shows what happens with an un/under-corrected lens.

LoCA is often corrected for with the use of aspherical elements in the lens design; which bend the light differently at different distances from the center... but the larger the lens/aperture diameter is, the more difficult it is to fully correct for the spherical aberrations (CA).

---

This next picture shows how using a smaller aperture increases sharpness, and reduces LoCA, by eliminating the edges of the lens where the light is bent the most (the narrow aperture light rays also exist in the wide aperture scenario; just not shown for simplicity/clarity).

Note that the light rays that are in-focus do not necessarily become any sharper/more in-focus when you stop down the aperture; they may actually/technically be less sharp (due to increased diffraction). But the combined total of all wavelengths is sharper, and that's what matters most. And you can see that the fringing is significantly reduced in the second image you posted; stopping down even farther would likely eliminate the CA entirely.

(Also note that the colors in this second diagram have nothing to do with wavelengths... the drawings were not originally intended to be presented together.)

---

So, yes... there is "something off" with the lens; it is not exceptionally well corrected for spherical aberrations/CA. But it is essentially a bokeh characteristic of that lens and not a "defect" as such. Pretty much any other f1.2 lens will show CA to some extent in certain situations; and almost certainly every other copy of that specific lens model will be essentially the same.

Answer 2

As a rule of thumb, the highest sharpness is obtained when you stop down 2 1/2 stops, about f/4. This is because, when the lens is at or near wide-open, in other words, light rays are traversing the peripheral of the lens. This is the area when the curve of the lens (figure) is the steepest. Additionally, the hyperfocal distance calculation is only an approximation. You should first test by setting the focus at infinity and shooting a series at different aperture settings. It would be advisable to run this test with the camera mounted on a tripod, this will rule out camera motion that could impair the test.