I disagree. True, SA, CA, and coma all constitute aberrations, yes...but what the lay-person refers to as "curved focal plane" while not technically or semantically precise, is nonetheless not an imagined phenomenon. Petzval field curvature does exist.
Taking SA as an example, imagine if the points in sharp focus can be joined in a 3D graph across the "focal 3D space" and your sensor can be magically bent to precisely pass through these points of maximal focus... It will not be a flat plane...while one could hope for a nice smooth bowl like graph, some complex optical systems, at least in theory, could generate nice central plane with concentric ripple with circular throughs and waves around this plane.
The challenge for optics designers is to address/correct SA, CA, coma and other spurious aberrations and distortions that arise in multi-element optics in a fair compromise and yet keep price and weight down...I do however agree that some SA was intentionally incorporated into the f/1.2 systems for artistic effect.
But field curvature does exist and is in fact the norm, to varying extent in both corrected and uncorrected optical systems.
So when the fanboys want corner-to-corner sharpness in the f/1.2 systems with zero CA and that melting bokeh, one has to just smile and be nice
Of course there will always be some however measurable
curvature to the focal plane in the real world. Is the Earth truly round? Is a flat wall ever truly flat
It was that very reason I took the time to qualified with "the 50L focal plane isn't curved or anymore curved than the average lens."
For the intent and purpose, the focal plane of the 50L is not abnormally nor excessively curved. In layman's term, it’s flat or of no real concern.
Spherical Aberration is not remotely the same as Chromatic Aberration nor does it have anything to do with focal plane curvature. It does however have everything to do with the shape/curvature of the spherical lens and can be corrected by way of aspherical contouring and/or counter elements (e.g. floating element).
The phenomenon of front focusing with the peripheral AF points is an entirely different matter and surely isn't called Focal Plane Curvature. Instead what was observed is none other than misjudgment by the phase-detect AF system in the presence of excessive Spherical Aberration. Despite such flaw, each peripheral AF points are still able to arrived at a given focus plane (even if not actually in-focus), but more importantly, be uniformly out-of-focus (aka flat).
On the other hand, given a lens that does
exhibit significant Focal Plane Curvature, one would be able to achieve perfect focus at any AF points (including peripherals), but the image would not be uniformly in-focus (only area near the center of focus). This phenomenon is independent of the type of focusing system used btw; AF or Manual. No amount of eye-balling MF will ever get you a flat pic. Clearly not what's going on here. We know the 50L is fully capable, especially MF'd or even Live-View contrast AF’d.
So what is it about SA that sends phase-detect AF to the funny farm?
Imagine at any given point on the actual in-focus plane, there exists a series of also in-focus points in-front and/or behind it (projected from different parts of the lens. see illustration). A typical phase-detection AF system simply doesn't have the smarts to make that judgment call. In a way, it did what it was designed to do, just didn't lock-on to the best one. This same mechanism is also the reason for Focus Shift at all AF points (even the center).
By stopping down (pinching of the aperture), you can cut down on the number of these false positives projections; less interference. For the 50L, the cut-off is somewhere around f/5.6-8.0 (DOF is large enough by then, that it’s tough to say for sure).
The bottom of the line is Canon had made a conscious decision to leave excessive (not little, but a lot) amount of SA uncorrected. Call it for the artistry or what have you, it’s there and it’s real – a real pain in the arse … until one learns to cope with it.
Through the ownership of the 50L, I have come to leverage the peripheral AF error against the center point focus shift.
AFMA wrt the center point is assumed. Focus shift at the center is some degree of back-focus when in the red-zone (within a several feet of subject, and f/1.2 > aperture > f/5.6)
. Knowing that switching from center to the peripheral points would result in some degree of front-focus, it could be used to negate the back-focus (from the focus shift). Examples:
- At wide open (f/1.2) I know I can shoot fairly reliable without thinking much.
- Inside the red-zone, I select one of the peripheral AF points best for the given aperture (you’ll have do some test runs for yourself to determine which pairs well).
- Outside of the red-zone, I shoot normally (any F-stop with any AF points).