I'm not talking about others when I reply to you. So don't try to involve others here.
1. You have been posting at least 200 post lately that only involves AF and focus issues. So it's nearly impossible not to read a few of those. And I know for sure you don't test or try your new lenses in a proper way for AF issues. You have described how you do in many posts. Handholding the camera and shooting at subjects like children or dogs. And you also say that those are perfect for testing the AF on a new lens. If I tested my lenses like that I would probably also have seven lenses with AF issues. You also test them at very close range with moving subjects. If you or the child/dog only move a couple of cm it will make the pic OOF. Many people here have told you it's not a good way to test the AF.
2. When somebody post about seven lenses and one camera with AF issues in a short time. It's very easy to suspect that the testing or the photographer when testing is doing something wrong. It's nearly impossible to buy that many different lenses in a rather short time with exactly the same problems or faults. (maybe you can if buying hundreds of lenses of course)
3. Everyone can get a bad lens with AF problems when buying a new lens. Even two if unlucky. But not seven + a camera in a short timespan.........
4 Test your next seven lenses with tripod and a static/still subject. Then I'm sure at least 5 or 6 will be good...............
5. Do you belive you are just unlucky every time you buy a new lens ?...Show more →
First of all, as I said earlier in this thread Canon has confirmed that my 5D3 has AF problems, which explains why I've had issues with several different lenses in recent months.
Prior to the 5D3, I had a 5D2 with Canon 35L, 85L, 50/1.4 and 100/2. I had no AF issues whatsoever. And yes, I was shooting my normal subjects at close distances and I never once had to "test" these lenses to make sure they focused properly. They just did.
With my 5D3, focus has been erratic and inconsistent with several lenses, which is why I suspected it had a problem in the first place. That said, the AF has been far worse with the two copies of the Sigma 35 I have than other Canon lenses like the 24-70, 50/1.4 and 40/2.8. It's probably not fair to judge the AF of a lens with a camera body that has AF problems; that's why I said I'll wait until getting the 5D3 back from Canon to make any judgments.
But the main point is this: I know how to shoot with wide aperture lenses, I have many fast lenses and previous camera bodies that I've never had any AF problems with. The same is true for other people in this thread. So why, then, if we notice problems with a new lens would it suddenly be an issue with our technique or the way we're testing a lens? That doesn't make any sense. There's a gap in your logic here.
This is a silly argument, especially in light of the fact that my 5D3 has AF problems. That is why I've had AF problems with several (3, not 7) lenses, not because of problems with my AF technique or the way I test lenses. I never had issues before with a different camera body, and I never found the need to "test" lenses on a tripod with focus charts. They just worked.
put about 150 clicks on this thing last night. after playing a bit i settled on +7 ma and it looks good. nailing focus at MFD and further away (15+ feet @ 1.4). outer cross type focus points on the 5d3 seem to perform just as well as the center.
AF is slower than canon L's i have... probably slightly slower than, for instance, the 135L, which isn't a rocket ship from MFD to infinity, but is good enough. it's a bit slower than that, but most of the travel is from 1-3 feet, so if you're constantly focusing on things that are a bit further away it's plenty quick.
i didn't have trouble locking focus in low light. it wasn't mind blowing but wasn't notably slower than my other lenses in these conditions. no idea about servo tracking yet. i'll have to get an event where i can test this out more. i usually use servo to track peoples eyes at f/1.4-2 with my 50mm when they're moving around a bit, so i'll be using this lens in the same way. i'll post how it is when i get a chance to really use it in that capacity.
macrobild wrote:
It is somewhat amusing to read all the comments.
First of all , you have to understand that Canon's AF has minor variations and AF accuracy is equivalent to f- 2.8 (3.4 to be exact) in cameras like 5dmk2 and therefore a AF test shall be made at F- 2.8 and with minor variations calculated in the results (and in the depth of field of 2,8). Depending on the lens and AF-motor the results can varies a lot.Take a series of 5 pictures and let the AF do a new measurement between every picture and you get maybe 2 of 5 really sharp. That a number of lenses exhibit the absolute focal plain with errors at f 1.4 and at short distance is therefore not surprising.
There are numbers of parameters who can make errors, AF does not work 100%
1.AF is wrongly calibrated in the camera
2.The lens is incorrectly calibrated,
3. Both the lens and camera are incorrectly calibrated...Show more →
Good points. Yes, AF is sort of a crap shoot. The lens/camera don't have an infinite number of focusing steps. Choosing the "best" is something of a gamble even with calibrated lens/body.
I recommend whoever solves the AF conundrum and brings us reliable and repeatable AF to receive the Nobel Prize.
cputeq wrote:
Or conversely, it says very little about his testing methodology.
A lot of the apparent focus problems with this lens seem to be closer-in, which may be closer than what a test bench would direct you to use or that quick focus tests would show.
Point is - just because Roger tested 30x copies and found them good doesn't mean they're all good at all distances - only the distance(s) he tested.
I'm not arguing for or against the lens, no horse in this race, but until the details of Roger's testing is known (have a link? I only have the one Sigma 35 blog post), then I would be hesitant to call his results iron-clad for the focus reliability. ...Show more →
Just to clear things up: the 30 copies tested are in a well lit lab and tested at 10 to 30 feet on stationary targets using center point AF. We do a back-frontfocus check using lens align, though, and all 30 copies were reasonably in spec, but again, well lit, 10-30 feet distance. In other words, no critical AF testing is done.
I leave critical AF testing to the reviewers.
My own copy shot in a variety of situations with a 5DIII was pretty good. I agree with what has been said -- maybe a tiny bit slower than the 35L but nothing significant.
macrobild wrote:
None have done a critical AF testing, it should take several weeks to include all parameters.
How do you know what all of us have done? I've done "enough" testing to know that my first copy was off in real world use; enough that I'm getting a 2nd one. At some point you stop the testing as you have conclusive evidence to determine that something is severely wrong.
cputeq wrote:
Or conversely, it says very little about his testing methodology....
Of course Roger was not doing AF testing, he was checking lens sharpness so his testing is not exhaustive by any means. The point however is that if this lens had any systematic AF issues as for example Sigma 50mm 1.4 had then there is a very good chance some issue would have turned up in his use of the 30 lenses. And we also have about 32 out of 38 users in this poll who report their copies work perfectly fine as well. So at this point there has been no evidence to suggest any specific AF issue with the lens other than the standard bad copy issue which happens with all lenses.
The 35mm looks like a winner and AF is good for me, but there's no question Sigma has a lot of history to live down and will need to prove itself to former customers who have been burned with flaky copies of the 50mm.
Shield wrote:
How do you know what all of us have done? I've done "enough" testing to know that my first copy was off in real world use; enough that I'm getting a 2nd one. At some point you stop the testing as you have conclusive evidence to determine that something is severely wrong.
what you have done I do not know, I replied that a proper evaluation of AF and put it in context would take weeks to do, for example, evaluate the 1DX vs D4 AF etc etc and I know none who have done that.
curious80 wrote:
Of course Roger was not doing AF testing, he was checking lens sharpness so his testing is not exhaustive by any means. The point however is that if this lens had any systematic AF issues as for example Sigma 50mm 1.4 had then there is a very good chance some issue would have turned up in his use of the 30 lenses. And we also have about 32 out of 38 users in this poll who report their copies work perfectly fine as well. So at this point there has been no evidence to suggest any specific AF issue with the lens other than the standard bad copy issue which happens with all lenses....Show more →
so it is, and its a good start to know
cputeq wrote:
Or conversely, it says very little about his testing methodology.
A lot of the apparent focus problems with this lens seem to be closer-in, which may be closer than what a test bench would direct you to use or that quick focus tests would show.
Point is - just because Roger tested 30x copies and found them good doesn't mean they're all good at all distances - only the distance(s) he tested.
I'm not arguing for or against the lens, no horse in this race, but until the details of Roger's testing is known (have a link? I only have the one Sigma 35 blog post), then I would be hesitant to call his results iron-clad for the focus reliability. ...Show more →
I would guess Rogers testing is better than the average FM member in this thread. He is also more used to do lens testing than the average member here. And how many other people have even tested 30 copies of this lens. And he say more about his testing methodology than 99% of the other testers in this thread
Sven Jeppesen wrote:
I would guess Rogers testing is better than the average FM member in this thread. He is also more used to do lens testing than the average member here. And how many other people have even tested 30 copies of this lens. And he say more about his testing methodology than 99% of the other testers in this thread
Well we now know a bit more on how he tests and he states no critical AF testing, which is what I suspected, as it would be entirely too much work to test every single lens at every single focal distance possible.
Plus, a lot of you guys are really missing the nuances of my previous post.
My caution was to not take Roger's word that 30 copies tested fine at whatever his standard testing ranges are (apparently 10-30 ft) as equivalent that those same copies would perform fine at ALL focal distances, and then interpolating that data to mean most Sigma 35s would perform flawlessly-- it could very well be Sigma's AF routines are lacking in close-focus situations (as some problems indicated here at FM seem to happen at those close distances), or it could be user error, typical QA variances, etc and we will hopefully get a 90-95% "good" rate with the lens.
Again I dont have a horse in the race, not a Sigma basher (100% agnostic, trust me), but to cite a respectable but still small test sample (performed at standard test bench ranges) and proclaim Sigma has suddenly cured all their AF woes is statistical folly. More time and testing required, but I am hoping this lens is spot-on, as I plan to buy it eventually.
