EDIT UPDATE 22.10.08. Please read my post on page two:

http://www.fredmiranda.com/forum/topic/698896/1#6299619





Hopefully people who have this lens can help sort out the AF problems encountered by most Canon FF users.

A quick recap written by Thrice in the Marmite thread:

1. The issue is back focusing when focusing beyond 10' on full frame bodies
2. Many copies focus perfectly on crop frame bodies
3. The more common issue (if there is one) on crop bodies is front focusing
4. Calibration is NOT the issue - calibrated full frame bodies matched to calibrated sigma 50/1.4's get the same issue. It is not correctable with lens profiles in mark III bodies.

Some more points:

1. Are there any problems with Nikons or is this a Canon-only issue?
2. Rumor: Lenses with serials higher than 1009xxx should be fixed. Can anybody confirm or dispute this...?
3. I have personally tried four different copies. All with bad AF.
4. All stores and importers seem to "not know anything about this problem". Great.
5. Please post comments or any knowledge to this issue. The lens gives fantastic images when focused...

For fun - a shot I made, using MF on the day I got my copy:

I had one with serial number 10088xx. I returned it because my Canon 50mm 1.8 was sharper at nearly all f/stops and distances. I was very disappointed.

Michael, I see from your profile you are an xt user. Do you think the sharpness problem was due to this systematic backfocus problem, which is the topic of the thread, or just general lack of sharpness in your copy? Did you have an opportunity to try it on a FF camera?

Good idea creating a thread with a more informative and descriptive title of what the marmite thread ended up being.

It was such a promising little (not really for a 50) lens, and after my two copies both having the exact same issue I was going to return for a refund and wait for the problem to be resolved.

However, both the AU distributor and a particular Sigma rep (named Vuslat) questioned everything from my camera, to my technique. As a result I won't be re-buying this lens or any other Sigma in the future, nor will I carry them in my store.

I will however, watch this thread with morbid interest to see if Sigma can bring this lens back from the dead.

It's a bit baffling why these manufacturers don't just release some
prototype copies to an army of beta testers and let them have at it.
The reward for thorough testing and reporting on the lens' performance
would be to keep the lens, if desired, or to get a coupon for a few hundred
dollars off any other product from that manufacturer.

Sigma could have saved themselves a lot of bad press by letting 10 or 20
of us put the 50/1.4 through its paces before they released it. It must be
a management problem.

I think this is a simple case of a very good optical design being mated to
a focusing system that isn't as good. Proper testing on Sigma's part could
have found the flaws right away. If they let a gang of users try it out
for them, it should surface right away.

Hear hear.

ovredal73 wrote:
Michael, I see from your profile you are an xt user. Do you think the sharpness problem was due to this systematic backfocus problem, which is the topic of the thread, or just general lack of sharpness in your copy? Did you have an opportunity to try it on a FF camera?

I don't really know to be honest. I'm not as adept in such matters as some. At first I suspected the backfocus problems but I think it just was just a general lack of sharpness. Tripod shots were just not as sharp overall, at all f/stops. Pitty! Dont have a FF camera... yet. Sorry I couldn't be more help.

i tested the AF abilities carefully. First with a 450D,
here.
Then i repeated the test with my next copy and a 5D and got the exact same result.

In short: both cameras were freshly calibrated by Canon. Both the lenses were calibrated to the cameras by Sigma Scandinavia. Sigma calibrates the lenses at a 40x focal length distance and in this case it means 2m (although the Siggy 50 is around 47mm only really).

The AF worked fine at 2m distance. At shorter distances I got a front focusing problem and at longer distances a back focusing problem. At very long distances it was a hit or miss (well, 2 out of 20 were a hit).

The image quality is way better than the EF50/1.4 and if the lens construction at least had permitted a smooth manual focusing operation I would have kept the darn thing. Now I'll wait for positive and confirmed good AF results before trying it again.

Cyberphoto (a well reputated Swedish on-line retailer) refused to answer my question about the return ratio. (They also didn't comment my next question which was me asking if I was allowed to ever by another Sigma from them...)

The test result from 3m with the 5D:


Cheers,

--
Jonas

Jonas, thank you very much for that test posting. Remind me not to waste time sending in my copy to Sigma Scandinavia...

You're welcome ovredal,

As you can see in the link I tried the AF in a systematic way and the results were disappointing, in a consistent way. I can accept that not all images are correctly focused but I can't accept having to stop down to f/5.6 to get things within DOF.

I went through the calibration routine as I'm living close to both a Canon Service Center (Mölndal/Göteborg, Sweden) and Sigma Scandinavia (Kungsbacka, about 20 km south of Göteborg).

After having done everything right and double checked it there was nothing else to do but sending also the second lens back. This is a pity. For the moment I'm using an EF50/1.4 instead but i would really like having a fast 50 with good bokeh.

Well, I tried. And btw, the serial numbers were 1001*** and 1002*** respectively. Here is still some hope for the future.

My thorough testing revealed similar results to what Jonas has reported, though
it does seem that at 3m distance mine does better. Mine is also serial number
1001xxx and I have tested it on a 5D. The EF 50/1.4 I have, by comparison, is
remarkably accurate on the same camera body. The region where mine starts to
show front focusing seems to be at 1.5m and closer. I do get significant backfocus
at long distances, such as 15m out to infinity, but since I don't shoot wide open
at long distance that much, I have decided to keep the lens for the time being
and enjoy its many good qualities. I was kind of obsessive about testing it
early on, but now I realize it has limitations and I work around them. If Sigma
eventually gets a fix for this issue, it will be a no-brainer first recommendation
for a fast 50 in this price class. I like the optical quality very much.

As I have mentioned numerous times elsewhere, my detailed AF results are here:
www.pbase.com/tswen/sig50af

Tim

Thanks for the post, Tim

ovredal73 wrote:
Michael, I see from your profile you are an xt user.

My brother's XT has terrible focus issues even using f/2.8 lenses, and it's been recalibrated once by Canon. It's rumored that the AF mechanism on Canon's low-end bodies is only accurate enough for f/4 and slower lenses.

I wouldn't even consider shooting narrow f/1.4 DOF lenses on these bodies.

Drew_Persson wrote:
ovredal73 wrote:
Michael, I see from your profile you are an xt user.

My brother's XT has terrible focus issues even using f/2.8 lenses, and it's been recalibrated once by Canon. It's rumored that the AF mechanism on Canon's low-end bodies is only accurate enough for f/4 and slower lenses.

I wouldn't even consider shooting narrow f/1.4 DOF lenses on these bodies.


i bought my canon 50 1.4 few years back when i was still using original rebel, its not focus as accurate as xxD or xD bodies but i have no problem with aperture 2.0 and smaller

I tried the Sigma on a 350D today and it focuses fine there. So now I have actually turned in my 5D for AF repair, to make sure that is not the problem. Maybe the FFcameras/5D are more sensitive to a weak AF precision level on the Sigma? Or does that sound like science fiction?

not science fiction, both of my faulty copies were fine on the 40D. I believe Sigma failed at reverse engineering canon's AF algorithm for a 50/1.4 lens on full frame.

why would the AF algorithm be different?

Don't know, I'm not a lens designer/programmer. Just saying this seems algorithmical as it only happens on full frame and only happens at 3m+. If it were an issue with the focus gearing design it would affect both APS-C canon and all nikon copies.

Did anybody try to focus while pushing the stop down button?

A possible explanation of such behavior is focus shift when stopping down. Its a common lens flaw

Andi Dietrich wrote:
Did anybody try to focus while pushing the stop down button?

A possible explanation of such behavior is focus shift when stopping down. Its a common lens flaw

Somebody please explain how Canon once again (5D MarkII) could place the DOF control button on the left side of the lens?! How the heck is one supposed to focus stopped down?

I tested the Siggy 50 with regards to focus shift and that is not the problem. See link in my first post.

I also think we mainly are discussing AF wide open here? I did all my tests with the lens at f/1.4 - and any possible focus shift problems are out of context.

(I did the same test with my 50/1.2L and Leica 80Lux (both sold) and had no problems identifying focus shift with those lenses.)

Out of curiosity...why would you want to focus stopped down? I've never had a desired to stop down to f/8 and focus at that aperture or anything. The only times I'll focus stopped down is on larger aperture alternative lenses, where I may focus my Takumar 50 f/1.4 at f/2 just to avoid the aperture action.

ovredal73 wrote:
I tried the Sigma on a 350D today and it focuses fine there. So now I have actually turned in my 5D for AF repair, to make sure that is not the problem. Maybe the FFcameras/5D are more sensitive to a weak AF precision level on the Sigma? Or does that sound like science fiction?

To clearify why I turned in the camera for possible AF repair before returning the lens - I found my 85L which I havent used in a year, to also (suddenly) have severe backfocus issues on the 5D, while not having it on the 350D. And to completely eliminate any question to the functioning on my 5D. In the repair shop log the camera is set to have AF adjustment done tomorrow, whatever that entails.

Could exclusively using old exotic MF lenses over the last year screw up my camera´s AF somehow?

thrice wrote:
not science fiction, both of my faulty copies were fine on the 40D. I believe Sigma failed at reverse engineering canon's AF algorithm for a 50/1.4 lens on full frame.

That is how I feel as well. However if my 85L also continues to focus badly, there is either something wrong with both those lenses or with the camera. So the only question now, is the performance of my 5D. Then I will start blaming the lens again.

André

Andre - make sure you're not testing under fluorescent lights. There are several of us (me included) who have bodies that backfocus under fluorescents with fast lenses (My 30 f/1.4 and 85 f/1.2 do it), but are dead on accurate under any other light.

Sorry Jonas, did not visit the link, just read what you folks have to say here.

Jman, only to see if there was any focus shift. There is an articel on focus shift on vanwalree.com > spherical