One_DaveT wrote:
(...)
Well I may just fall into that category. When I initially reported, I had been mostly shooting in the < 10 foot range with just a small few beyond. As I reported, these seemed fine. Having read the responses, I did several rounds of distance test shooting (10,20,30,40 feet). I found many of these to have a back focus aproximately equal to the distance of the DOF at f1.4.
Regarding serial number, Yes, I meant 1007***.
Putting this into perspective, this particular use case which may have an issue is an odd one for my style of shooting. I don't usually shoot at a distance with a 50mm, and when I do for a landscape, I would likely have the lens stopped down a bit. If I'm trying to close down the DOF, blurring out the foreground and background, something with a longer focal length like a 135/2 or 200/2.8 might be better suited.... just my preference. ...Show more →
Thanks again. Then we are down to two or three (maybe) working lenses again.
I agree that using a lens wide open at long distances is uncommon. But 3m isn't that far away and I can think of groups of people and street shooting where the AF at f/1.4 need to do better.
I wanted the water not too blurry, nor frozen, the trees in the background blurred and still possible to handhold. Yes, I sometimes need f/1.4 also at longer distances. Now I could focus using Live View. I wish the tress were one notch more blurred but I guess I can fix that in another round PP. Next tiem I hope to have a FF camera and a fast 85 or 100mm lens instead. As always it is about doing the best with what's available.
I have been watching this thread for some time, and with some trepidation went ahead and got a new sigma for my canon 5d. My lens is a 10089** and after a day of informal testing (walking around shooting near and far) I get very acceptable results with little or no focus error. These results are at least as good as my other lenses and I am very pleased (so far) The lens is also amazingly sharp especially stopped down to 1.8, so hopefully more good lenses will get out there.
I just got this lens, and after two days, it is clear:
1) most of the time, it is front focusing
2) the "clarity/contrast" at f/1.8 is worse than my canon 50MM f/1.8 mkII
BTW, my serial # is 10017xx
Here is an example of the issue I'm seeing:
Canon 40D mounted on a tripod
- The piece of heavy printed card board is mounted at an angle,
roughly 45 degrees from horizontal
- settings are 400ISO, 1/125 speed, f/1.8
- center focus point on the word "toddler", specifically between the two d's.
- 2 second delay for shutter actuation to avoid any shake
autofocus.jpg shows that the word "infant" is in focus (it is less
pronounced in this resized version, but still visible. The word
infant is "in front" of the word "toddler" due to the angle of the
card board.
manual.jpg is the same picture taken with manual focus and live view
magnified 10x. Here, you can see "toddler" is clearly in focus.
Finally, here is the same setup, auto focussed with canon 50mm f/1.8:
manual focused canon 50mm f/1.8:
I'm a little bummed about this. The build quality is exceptional, the full time manual focus is a big plus, I really wanted this lens to kill the MKII.
You might consider trying another one. But I have found some degree of
front focus on mine at close focusing distance. But it's not a big amount.
It's a great piece of glass when properly focused. I wish Sigma had totally
nailed the AF, but the more I test it the more I see that they left some
design work undone. Still, don't write this lens off right away. It's worth
obtaining a good copy and seeing if you can live with its limitations.
Check out my results here if you are interested. I'm adding more results right now:
Glassbottle wrote:
I wonder, do people find the same thing at a distance and wide open when using other lenses with wafer-thin DOF such as the 50L and the 85L?
After doing more extensive, controlled testing at distances around 20 to 40 feet,
I can see that my Sigma tends to backfocus more than my Canon 50/1.4
though the Canon is not perfect either. I've determined I can live with it, but
the Sigma does appear to have more trouble nailing the focus at long distance,
including at infinity. I would think this is quite fixable by Sigma, but it would
require more than a simple linear calibration adjustment, as the Sigma tends
to also front focus just a bit at close distances. So, it appears their
correlation between focus distance and motor rotation has the wrong slope,
so to speak. I had to pixel peep a lot to really accept this, but I'm still enjoying
the lens and starting to see some workarounds to this issue as well.
I'm desparately trying to finish loading up my autofocus test results here:
Tim, thanks for the link. Based on your findings, it seems that Canon 50 f/1.4 is the clear winner in the tests that you ran. It is also about $150 cheaper, so I think Im going to return the sigma and pick up the canon 50 f/1.4 (or just keep m f/1.8-- although i had to send it in to canon for service recently). It's really too bad, I was hoping this lens would be a great portrait performer.
Oh, I think it would be a great portrait performer on a crop body, but there is not much
reason to go for it if you never shoot below f/2.8 as the Canon 50mm lenses do very well
when stopped down. The reason you would go with the Sigma would be for excellent
optical performance at large aperture.
The reason I did all of those detailed tests was
1) to see if all of the complaining on the boards about the focus was valid [it is partially, but
the extent of its effect on real shooting will be quite dependent on your shooting habits];
2) to get to know the ins and outs of my copy to know how to work around any problems and
maximize its strengths. [To me it's worth it to find work arounds for its focus limitations, because
it is optically stunning wide open.]
But I would not recommend the lens without some reservations. I fully understand why some will
not find it worth working around.
Af f/1.8, the canon 50mm MKII out performs the sigma at f/1.8 based on my test shots, as well as real world shots. Also, at those small DOF values, the focus accurracy is crucial, and the sigma is failing there as well.
Glassbottle wrote:
I wonder, do people find the same thing at a distance and wide open when using other lenses with wafer-thin DOF such as the 50L and the 85L?
I think this is a fair question, and one that should have been answered first. I went out this afternoon and did some test shots 10,20,30,40 feet on 85LmII, 35L and Sig 50 all at f1.4. The results were a little surprising to me. Sample size was small, but this is what I got:
85L
10 ok
20 - slight back focus
30 - slight back focus
40 - slight back focus
Sig50
10 feet back focus
20 ok,
30, ok
40 ok
35L
All were ok, but the DOF is quite large.
Shooting at these distance at f1.4, is still an odd use case for me, and likely why I never noticed my 85L off like it was. I think it would be good when testing to keep in perspective the margin of error that other lenses these might be compared to do. Rather than now saying that my Sigma 50mm has a focusing problem at distance, it might be fair to characterize it as, my Sigma 50/1.4 focuses better at distance than my 85L for an unusual test case that I rarely shoot at anyway ;-)
In the interest of complete disclosure, I'll say that looking over my test shots from a couple days earlier under the same conditions, the Sig50 was right on at 10& 20 feet, and had back focus at 30&40. This tells me that sample size may play a more significant role for accurate characterization.
changes13 wrote:
Af f/1.8, the canon 50mm MKII out performs the sigma at f/1.8 based on my test shots, as well as real world shots. Also, at those small DOF values, the focus accurracy is crucial, and the sigma is failing there as well.
Not sure whether you mean the 50L MKII or the 50/1.8 MKII at f1.8, but I thought I'd offer up this handy comparisons to the Sig 50/1.4. Mouse over the arrow in the upper middle to see the image flip to the other lens.
One_DaveT wrote:
Not sure whether you mean the 50L MKII or the 50/1.8 MKII at f1.8, but I thought I'd offer up this handy comparisons to the Sig 50/1.4. Mouse over the arrow in the upper middle to see the image flip to the other lens.
I enjoy the-digital-picture.com comparisons, and find them very useful, but you need to be careful interpreting them with the 50/1.2 (mII ). If you read the review of the lens, it's admitted that the method used to shoot the test targets is basically showing the 50/1.2 at it's weakest...up close and stopping down. If this is how you intend to use the lens, then fine the results matter, but if you typically shoot the 50/1.2 at longer distances the test doesn't really give a good indication of what the lens is capable of delivering.
Just a quick update, I too saw a little front focusing close in, but the focusing was very accurate after a few feet. There were a few out of focus images that I couldn't figure out but on the whole I really loved the lens. Unfortunately, the focusing ring started to squeak, getting louder with use so I returned it and am expecting a replacement soon.
Also this lens blew me away stopped down to f 2.8.
To be clear, my other 50mm is the el cheapo nifty fifty, non L lens. I did not bother to test focus at greater than 8 feet because I intend to use this mainly for portraits. When I first got the lens, it was clear that the portrait pictures we were taking were out of focus, sometimes wildly. After noticing that, I started doing more "controlled" tests and determined my copy pretty much always front focuses between 2 and 8 feet, even stopped down to f/2.5. I did get one spectacular shot at f/1.8, but it was the exception rather than the rule.
One thing I noticed on my copy and am seeing on a lot of sample shots is longitudinal CA on the OOF areas, usually green in the background and purple in the foreground. Anyone else noticing this?
I've read that this issue isn't as apparent in the focused image if you manually focus (since AF system more sensitive to green it focuses green light and purple is OOF or something similar) but in the out of focus areas it simply means the lens is farther from apochromatic than normal.
DPReview wrote:
it's important to realise that depth of field is so limited at these apertures that even getting the subject in sharp focus is a trial, let alone the corners (even a slight relative movement of the photographer and subject will result in a misfocused image)
). If you read the review of the lens, it's admitted that the method used to shoot the test targets is basically showing the 50/1.2 at it's weakest...up close and stopping down. If this is how you intend to use the lens, then fine the results matter, but if you typically shoot the 50/1.2 at longer distances the test doesn't really give a good indication of what the lens is capable of delivering.