It is indeed difficult when regular back focusing is taken to account.
My experience has been that I can correct the BF via MFA and it would be as spot on as it could be at f1.2 and OK between f1.2-f1.8 perhaps, but then between f2-f5.6 the focus would shift towards the rear again. I am not entirely sure what that proves, but IMO that's the focus shift at work since it's OK wide open. No?
But no doubt, part of the problem is the regular BF. There seems to be a lot of copies out there with that. What's also confounding is that Canon doesn't seem to be able to precisely calibrate these lenses.
Galibier wrote:
Actually, I think depth of field should be split almost 50/50 at these distances, with slightly greater DOF to the rear. J. Allen's 50/1.2 exhibits what I think of as the standard dose of focus shift with this lens. To "fix" the focus shift would seem to require either a re-design, or firmware to re-adjust focus depending on aperture and subject distance, as Konica did some years ago. I wonder if the reported widely-varying focus performance of this lens has discouraged Canon from tackling this issue, or if they just don't consider it worth the effort.
Nick Nishizaka wrote:
What's also confounding is that Canon doesn't seem to be able to precisely calibrate these lenses.
Sorry for topic drift here, but I think Nick's observation is an issue of service workmanship. I don't believe it's exclusive to the 50L. Over the last few years I've had what seems to me a disproportionate number of L lenses < 135mm needing service, maybe because they've all been purchased used. At any rate, some of the lenses needed two trips (and more) to be serviced satisfactorily. I do wonder about the training, consistency, and competency of the technicians.
Shuko wrote:
From a bit over 4 feet away, so somewhat decent shooting distance instead of just doing it at minimum focusing distance to really bring it out.
Nick Nishizaka wrote:
But no doubt, part of the problem is the regular BF. There seems to be a lot of copies out there with that. What's also confounding is that Canon doesn't seem to be able to precisely calibrate these lenses.
I also find it confounding.
Forgive me for repeating myself but I struggled with this some time ago, with several copies and calibration attempts. In the end I found that AF could be different depending on the light. See the figure below where the difference between incandescent and fluorescent light is illustrated. In fluorescent light as well as daylight I have BF (right sample). In incandescent light (left sample) the DOF is pretty centered.
So my theory is that Canon calibrates the lens in their special calibration light and everything looks perfect, but in daylight it BF because the AF system makes a slightly different judgement in this light.
I could observe this phenomenon also with my 24L but the important difference is that this lens obtains best focus in daylight, so it front focuses slightly in incandescent light instead. So this theory also assumes that the 50L is playing some trick with the calibration light that other fast lenses don't do.
Both the 50L and the 85L have quite nasty lateral CA, so light temperature will affect focus accuracy. All CCD/CMOS devices have a very high sensitivity to IR, and to balance this there's an IR-filter in the AF lens/mirror assembly. Camera makers seem to want to keep as much low-frequency light as possible though, cutting away to much would seriously affect the AF-capabilities in low-light conditions. Just from the top of my head the figure 1.5% difference in focal length (within the visible light spectrum) appears when regarding the 85L, and I thought the 50L was even worse from the sample that I've used.
But the focus shift has nothing to do with AF, to test the focus shift you should be in MF! No focus adjustments between the test shots at all, just the stopping down. Isolate the components you want to test, that's a basic rule for objective testing.
At 1.5m distance, the copy I tried shifted so that around F/2 parts of the target i shot that were "in focus" at 1.2 were in stead "in front of focus" - ie, less sharp than at F/1.2, and this is what "Focus shift" means...
The best setup for the 50L seems to be when the camera/lens is set up to have a slight front focus at wide open. That will keep shots taken in F/2-F/2.8 centered in the focus field.
Interesting info Suede. I don’t have the knowledge to make full sense of what you wrote, but it seems very plausible to me that things that you mentioned are playing a part in the regular back focus that is reported once in a while. Because it is regular BF that I am talking about. The two samples that I posted are both wide open, and the right one, with BF, is what I live with. So when adding the focus shift on top of this, it just means that I am out of focus even at f/2.8, which is hard to swallow. Currently I am using workarounds to counter it.
I agree completely that the lens should not be refocused when testing the focus shift. My experince of the shift is that the front end of DOF is about the same at 2.8 as at 1.2.