Ok, I shifted gears here to try something I had not done before. I had read on Castleman that as you moved away from minimum focus range the back-focusing problem resolved. I chose to shoot at 1.5m on Coke bottles. Again the target is the second bottle from the left. We start at f/1.2
For fairness I included the 50 f/1.4 shot at f/2.8. It also back focused some on this target so this may not be an easy target to hit. I dont know why...
I then thought this may just be a difficult focusing scenario that I have inadvertantly created. I wonder if the 85L could resolve these bottles at 1.5m. First shot 85L at f/1.2...
Thanks for doing the tests! Looks lots like what I have seen from all the copies I have tested also. (5 of them tested now)
I was in Dallas earlier this week and a local store there had a UV0101 and UU1000 lens that they let me test. I set up a battery test along with a high contrast box target and used center point AF (the center point AF was targeted on the Black portion of the Canon Close Up Lens box). Here is an f/2.8 shot (cropped) from the UV0101 lens.
The third battery from the left was aligned exactly with the face of the Canon Close Up Lens box. As you can see, the lettering on the second battery from the left was much closer to being in focus. It was approx 1" behind the third battery and the focus target. The total distance from the target for this shot was about 30 inches...
And yep....I'm waiting for what I hope will be a firmware fix before I buy another lens...
After I totally disgusted myself with these results, I asked myself I wonder what the 50L would do on my XT? Boy was I surprised...
Focus chart at f/4.0
I guess if you are having trouble with your 50L you just need to get an XT!!
After wracking my brain it seems that the 50L has a problem focusing using the precision points (ie center point on 5D) which sensitive to f/2.8 and faster lenses. The minor points are sensitive to f5.6. To my knowledge the XT only has non-precision f/5.6 type points. This seems very likely a software issue and maybe could be fixed with a firmware adjustment to the body or lens. I will post a few more conclusions later
I hope I got all the photos in the right holes. If you spot an error, please let me know.
ward1066 wrote:
which minor point are you using? The one to just the right or left of center? and are you still using that point centered on the battery when you AF? thanks
Iam using the lower point (below the center point) which is centered on the battery. But you can use any of the minor points on the 5D and get these results assuming the point you use is centered on the target.
More Conclusions... and Questions
1. The 50L is optically a superior lens. I believe that it has a software issue in focusing. The issue seems to be worse when the precision focus points are used. If the lens is manually focused it is on par with the 85L and other L primes and is superior to the 50 f/1.4.
2. The problem seems worse on higher end bodies like 5Ds and 1 - series. This may be due to more precision focus points or different interactions with the software in these bodies. Has anyone tried the lens on a 20D or 30D?
3. I suspect that Castleman's results showing the 50 f/1.4 to be superior to the 50L above ~f 2.4 or so have been affected by the focusing issue. I will let Mr. Castleman be the final judge of whether Iam right on that one.
4. The problem seems to get worse with increasing numerical aperatures. Maybe someone who knows more about optics than me can explain this.
5. In my test the problem was mild at ~0.5m but got much worse at 1.5m. It was some better but not entirely resolved at ~2m. This would seem to spoil many real world shots. I would like to know if the problem ever completely disappears with increasing distance to target. I need a new test with bigger targets to test this theory.
6. In real world shooting I get a surprising number of keepers given the obvious systematic problem with the lens. Are there so many variables in a real world shooting that we can't control that this focusing problem gets lost part of the time?
I got tired of worrying about the hit and miss AF at near distances and bought a EC-S focus screen I think it works really well in low light and off center compositions.I guess this will hold me over until Canon figures out what to do, if they do.
mbailey wrote:
.... If the lens is manually focused it is on par with the 85L and other L primes and is superior to the 50 f/1.4.
Has anyone demonstrated this yet? A manually focussed shot at 2.8 and 4.0, from a few feet, and sharper than an identical set up with the 50f1.4? I haven't seen that result yet. A lot of testing has demonstrated back focus in this range, yes, but I haven't seen a super shot done manually in the range in question, to once and for all show that the optics in this range can deliver if AF is taken out of the equation.
..
I suspect that Castleman's results showing the 50 f/1.4 to be superior to the 50L above ~f 2.4 or so have been affected by the focusing issue. I will let Mr. Castleman be the final judge of whether I am right on that one.
Yes, I am waiting with curiosity for his reevaluation of this.
RoyPertchi wrote:
Has anyone demonstrated this yet? A manually focussed shot at 2.8 and 4.0, from a few feet, and sharper than an identical set up with the 50f1.4? I haven't seen that result yet. A lot of testing has demonstrated back focus in this range, yes, but I haven't seen a super shot done manually in the range in question, to once and for all show that the optics in this range can deliver if AF is taken out of the equation.
Yes, I am waiting with curiosity for his reevaluation of this.
Here is a manually focus shot from 50L at f/2.8. This was not posted earlier
ward1066 wrote:
I got tired of worrying about the hit and miss AF at near distances and bought a EC-S focus screen I think it works really well in low light and off center compositions.I guess this will hold me over until Canon figures out what to do, if they do.
my faithful practice subject
FWIW, I use an EC-S in my 1D2N, which may account for the lack of missed targets during my own assessments, giving me the false impression that my lens was fine.
Mike, a most excellent write up. And I was more than happy to send my $1600 lens half way across the country for the greater good.
Having seen two copies compared side-by-side, we have found that the missed focusing problem is consistent with earlier reports. At least we now know that the 50Ls are built to the same specs! I think we should collectively, as a community, make our voices heard to Canon. I know a lot of us are past our return period and are stuck with what we have. So it's either Canon addresses the issue and fixes it, or we're stuck with a lens that doesn't perform optimally at every level.
To add, I do believe that a firmware fix is in order to address the focusing issue. We've already proven that the lens is great optically. Being that the lens performed flawlessy on the XT and subpar on the 5D and 1D series tells me that it's all in the software.
I think we should collectively, as a community, make our voices heard to Canon. I know a lot of us are past our return period and are stuck with what we have. So it's either Canon addresses the issue and fixes it, or we're stuck with a lens that doesn't perform optimally at every level.