I went back and tried to follow autofocus issues as reported in this thread. Six people have reported fairly serious autofocus issues: Simon_k, hubsand, 05xrunner, Jonas B, badlydrawnboy, and thrice. At least four of these report really bad back focus at distances longer than 10 feet. Trumpet-guy has sent his for autofocus review and it will be great to hear back what is found. Four others haven't reported on their autofocus or at least I didn't find their reports: MichaD, perspective, martsmith42, Kjetlis. Chez_Wimpy is the only one who has reported that his autofocus works well. I apologize if I missed anybody else who posted. I would love to hear from those who haven't reported on autofocus on their lens yet. Right now it looks to me that at least a lot of the lenses have a serious autofocus issue. Let's hope it can be fixed. Perhaps Chez somehow hit the lottery, but he also may have found a work around. He reports always using AiServo and center point only. Thrice's shots looks a little better with AiServo too, so maybe this helps and if Chez's lens is a little closer to being in spec fixes issues. It will be interesting to see how all this works out. I hope people keep reporting here.
As I understand people here (didnt follow the thread), this is normal back/frontfocusing which is a calibration issue and not necessarly a design flaw.
Focus shift occurs when the lens changes the focus when used at smaller apertures.
This seems to be the problem with the 50L. The AF mesures the distance wide open and the focus shifts when your camera closes the aperture of the lens.
Andi, the focus seems to be correct at close distance and way off at greater distances.
I don't think that's not normal backfocus. it should be off at any distance or at least get better at distance because of the increased DOF.
Or am I thinking the wrong way? I never had a lens that focused wrong.
Jul 12, 2008 at 01:57 PM
Steve Spencer Offline • • • • • • • Upload & Sell: On
I think it is still an open question what is causing the autofocus problems reported here. For at least four of the people it seems to be massive backfocus--missing focus sometimes by two feet--from longer distances but pretty good focus from close range. It will be interesting to see if calibration can fix these problems. I agree it seems different from the focus shift problems reported with the 50L, but whether it is a quality control problem or a design issue to me at least remains to be seen.
Yes, for me it is not a normal focus shift problem as it occurs at the widest aperture (not when stopping down as Andi describes).
I am also very curious to know whether this can be fixed with calibration or whether it's a design issue. That's not a question Sigma will answer over the phone or via email, though. The problem is that if I send it in for calibration and they then determine it's a design flaw, or they can't fix it, or they say they fixed it but it really isn't fixed (all of which are possible) then I miss my return with B&H and I have a defective lens on my hands.
I will probably return mine and then wait to see how this pans out.
Thanks to Steve Spencer for keeping track of who has posted their results with the lens here. Of course there are many more people who have not posted here or elsewhere who 1) don't have focus problems or, 2) haven't realized they have a focus problem because they haven't shot anything at a considerable distance and relatively wide aperture.
While I didn't do any real focus tests I haven't observed any focus errors. Neither near or far.
The AF feels a lot less wanky than the one of my Canon 50/1.4. Which is to say the Canon is really horrible. I'm not really an AF guy anyway so a real plus is that you can actually manually focus the Sigma which you cannot really say of the Canon. It's obviously not as good as a nice MF lens in this regard but it'll do most of the time.
As you move further away from a subject, the subject becomes smaller relative to the whole frame and thus relative to your AF sensors. The AF sensors are larger than the lights in the viewfinder would suggest. I think the camera is just focusing on the contrasty area behind your intended targets.
In the shots of the stereo, the cables behind the stereo are in perfect focus and they would still be in the AF area if those are 100% crops. The same applies for the flag in front of the building in simon_k's post and in thrice's distance test with the paint can.
I don't recall having major focus problems at distances over 10 feet.
But as I mentioned earlier, I have had so much inconsistency with
Canon autofocus in general that I rely on manual tweaking wherever
possible, so I haven't done any tests with purely autofocus alone.
I do think that the infinity focus was pretty close to right on, with my
copy. I know this whole autofocus question is annoying, but I suspect most
shooters could still make great use of this lens, even if they need to be
careful with the focus at large apertures and large distance.
We'll have to wait a little more, but we will continue to see focus
test reports coming in. I expect joe mama will report on my lens within
the next two weeks (his 5D died and is getting repaired now), and I will
do an AF check when I get it back.
asabet wrote:
Thrice, sorry to say, that looks like AF hell.
From thrice's example one page back, I personally would never expect AF to work within that threshold (table in front of that background a room away... not in my experience). That is way beyond the tolerances I expect.
asabet wrote:
C.W., I know how you feel about the Sigma 30 and the Sigma 50, that people find problems because they go in looking for problems, etc. In that case, please consider selling your copy of the Sigma to me at a small premium and then buying another copy for yourself.
With my thinking, I have no doubt people with "problems" would find them with my lens as well - even with a big placebo "Chez Wimpy Certified" mark.
In terms of BF at distances, there was one specific set of shots, maybe 3 or 4, where I was shooting about 40-50 feet out at f1.4 in the gym where I missed the AF target completely. The back wall (covered by a contrasty metal grill) took the focus, but it was obvious to work out why. That behavior is well within my typical 20D focus error range, so I thought nothing of it.
There is no magical "10feet onwards" boundary on my lens, if anything it performs more reliably at those distances (the 30/1.4 got into a danger zone about 10-15feet out), and I have tons of shots in the 10-50 foot range at f1.4 outdoors on my flickr sample gallery for anyone that wants free-account restrained downsized images (so you will have to trust me that the focus at 100% is not obviously off!).
As for focus shift, can't say I have detected any at all. The battery focus test I did showed an even front/back dispersal as I stopped down. The "landscapes" were done in rapid succession, hand held (bad insects and no repellent!), with a single AF read on the houses out there near infinity. MF would have probably done better (but no live view on my 20D!)
BTW, my serial is 1002905... and was from day 1 in Japan (June 14th). People got OLDER lenses overseas... that is very strange.
Edited by Chez Wimpy on Jul 13, 2008 at 03:42 PM GMT
Well I must have gotten a lemon because I wasn't at all happy with its AF performance. If you used MF it was sharp wide open and very good at any distance but when using AF my focus plane was all over the place depending on what distance you were to the subject.
After testing it on a tripod, MLU, and cable release and getting inconsistent and inaccurate results focusing on varying distances and subjects it was packed up and sent back to the dealer. I even gave up on testing and tried it on everyday situations and the results were even worse. Tested on a 40D, 1DMKIII, and 1DsMKII.
Too bad because this would be a very nice piece of glass if the AF performed like my 35 f/4L. I would have sent it to Sigma for calibration if I knew for sure that was all it needed but if it is a design flaw that can't be corrected thats another story. Using MF the lens shows promise in sharpness, contrast, and color but the AF accuracy was lacking.
Sam N wrote:
As you move further away from a subject, the subject becomes smaller relative to the whole frame and thus relative to your AF sensors. The AF sensors are larger than the lights in the viewfinder would suggest. I think the camera is just focusing on the contrasty area behind your intended targets.
This is definitely a possibility focusing on small objects but I used large objects at a distance over 10 ft. for testing and the results were the same. I was aware of the possibility on the focus point overlapping and picking up the BG so I made sure the subject was large in the frame when shooting at a distance.
One subject was a brick mailbox that stood about 4ft. high on the edge of a residential street and took up 1/4 of the frame. It was about 15 ft. away and the lens focused on the far curb of the street. I used a tripod, MLU, and a cable release and tried it on a 40D, 1DMKIII, and 1DsMKII. All three of these cameras function perfectly on all of my lenses.
Anything inside of about 3 ft. focus fairly accurately but as you moved beyond that 3 ft. mark the results diminished. I tried using MA on my 1DMKIII to compensate but it went beyond the range for MA. Using MF though resulted in some very good results. This lens was sharp wide open at any distance using MF.
Chez Wimpy wrote:
BTW, my serial is 1002905... and was from day 1 in Japan (June 14th). People got OLDER lenses overseas... that is very strange.
Edited by Chez Wimpy on Jul 13, 2008 at 03:42 PM GMT
Maybe not so strange after all. Maybe Sigma wanted to release this
lens at about the same time all over the world, so the earlier production
was put on the boat and sent to Europe/America. The latest production
was sent to the Japanese distributors. Just a theory anyway.
trumpet_guy wrote
Maybe not so strange after all. Maybe Sigma wanted to release this
lens at about the same time all over the world, so the earlier production
was put on the boat and sent to Europe/America. The latest production
was sent to the Japanese distributors. Just a theory anyway.
Even with the early batches overseas, there was still a two-week + delay between the Japan and elsewhere release... I guess shipping takes longer than I thought - or alternately they favor the home market with the more polished goods (going by the Japanese online reviews so far, no complaints of BF locally)
Sam N wrote:
As you move further away from a subject, the subject becomes smaller relative to the whole frame and thus relative to your AF sensors. The AF sensors are larger than the lights in the viewfinder would suggest. I think the camera is just focusing on the contrasty area behind your intended targets.
In the shots of the stereo, the cables behind the stereo are in perfect focus and they would still be in the AF area if those are 100% crops. The same applies for the flag in front of the building in simon_k's post and in thrice's distance test with the paint can....Show more →
thought so too.
but look at this - I couldn't find any bigger subject.
Full frame Canon 50/1.8 mk2 @f/1.8
100% crop
Full frame Sigma 50/1.4 EX @f/1.8
crop 100%
The purple cast of lateral chromatic abberations indicates that the focus is way behind the subject.Both pictures were taken from a tripod
Each of your tested shots is taken below 1/50 exposure time. These test shots therefore seem unable to proof anything. I'm sure some of you experience backfocus, but I haven't seen any reliable proof of this I'm Afraid. Also, when focusing on a small spray can from so far away it's to be expected that the AF focuses past it.
The amount of complaints about sigma's AF does seem to indicate a problem, but from these pictures I can't tell whats wrong (apart from the exposure times obviously)
thrice wrote:
Took it out for a quick shoot, and even with me compensating for the back focus by MF after one-shot it's so hard in the dark coupled with such a terrible distance scale.
I'm likely to do the same as badlydrawnboy as I'm in Australia and the distributor here won't touch imported Sigma lenses so I may have to send it back to B&H... also reluctantly.
@biotar: I'm pretty sure he used a tripod. nobaody could hold a 2sec exposure like in picture #4.
My shots were all taken at SS far above 1/500s (ran into 1/8000 limit sometimes) AND from a tripod.
I tested it twice to be sure it's no user error.
Very sad because the lens can produce outstanding results.
Yeah I used a tripod I have a vague idea what I'm doing. Every shot was on a tripod, and the camera indicated the spraycan as the activated focus point using the centre focus point activated and the secondary (invisible) focus points disabled in custom functions. Hang on one moment and I'll do a definitive test for you, also on a tripod.
A colleague at work has a friend at the repair centre in my city who can get my lens calibrated, not cheap, but much cheaper than a new lens. Incidentally the guy @ the repair centre has one on loan and is having issues focusing closer than 10' on the EOS 20D, he is an accredited and qualified professional photographer.
One shot at 5' and one at just over 10', no issues ensuring that the milo can dominates the central area and the camera definitely thinks that's what it has focused on.
Pretty obvious if you ask me.
Make no mistake I love the lens, to the point where I'm paying to have it calibrated so that I don't have to send it back. I am simply sharing my findings and trying to establish if it is a common trend.
And in a vain effort to apologise for the the negative thread derailment here is some marmite bokeh to show front/back bokeh of this stellar lens
