Sigma 18-35 Art inaccurate with 80D DPAF

Hi,

I just wanted to comment on this old thread with the same subjectbut it has reached the maximum number of allowed posts. No more to say!

Three days ago I decided to purchase the Sigma 18-35 despite this thread. I am loving my Sigma 50-100 and hoped Sigma has already resolved this issue with inaccurate DPAF on newer Canon cameras. It seems they have not.

The first copy had severe backfocus, both on my 80D as well on my M5. I brought it back to the store and tested another copy (directly at the store). Same thing. It was way better than the first copy but still not accurate enough.

I am really sad, I'd love to use this (optically) great lens on my two bodies. But without working AF its use would be too limited for me.

Here are some test shots I did with the first copy. Not very scientific, but hopefully usefull:

Shooting distance about 1,5m. I had no decent light available. So shutter speed was only 1/50s but on a tripod. I tried 18mm and 35mm each with manual focus and DPAF/Liveview. I did not test with PDAF because I didn't calibrated with the dock yet (and it would not have improved DPAF performance anyway).

Here is always a 100% crop of the sharpest image from a series of 3 test shots each:

18mm, MF:

Sigma 18-35 at 18mm/1.8, manual focus on 80D

This is how sharp this copy gets at 18mm.

18mm, DPAF:

Sigma 18-35 at 18mm/1.8, DPAF on 80D

Severe backfocus, way beyond the maximal scale of 6cm.

35mm, MF:

Sigma 18-35 at 35mm/1.8, manual focus on 80D

This is how sharp this copy gets at 35mm.

35mm, DPAF:

Sigma 18-35 at 35mm/1.8, DPAF on 80D

Again severe backfocus, this time only somewhere behind 6cm, maybe 8cm.

And now, for comparison the Canon 35mm f/2 IS with a similar setup:

35mm, MF:

Canon 35mm f/2 IS at 2.0, manual focus on 80D

35mm, DPAF:

Canon 35mm f/2 IS at 2.0, DPAF on 80D

Interestingly the Canon backfocuses at well with DPAF. Sharpest position is about 5/6cm behind the target, I'd say.

This makes me think. Is may setup wrong? I tried to use DPAF always with the full focusing box on the target. Some say that DPAF uses an average of all pixels in the box.

Maybe DPAF is not that accurate as Canon suggests?

DPAF in liveview doesn't use or benefit MFA, if I have done my research correctly.

Also it is possible that DPAF is not fully supported in your sigma lens.

damage98 wrote:

DPAF in liveview doesn't use or benefit MFA, if I have done my research correctly.

Also it is possible that DPAF is not fully supported in your sigma lens.

Exactly, that's the point. With MFA you can only adjust any discrepancy between location of sensor and PDAF (phase detenction) sensors. Same with Sigma's USB dock.

With DPAF (dual pixel in the normal sensor) there should not be any need to do adjustments in the body or in the lens. However, the Sigma 18-35 was developed before Canon introduced their cameras with DPAF. And so far Sigma failed to make it work.

Although it seems there are 18-35 lenses out there that work with DPAF. Or their owners just don't use it that much in liveview mode!?

I just repeated my test from the original post with my Sigma 50-100 and this lens works great as always.

MF: Best results

PDAF: Quite good results (I didt thorough calibration with the dock last year)

DPAF: Okay results (but not as sharp as PDAF with calibration)

It seems that even with this generation of lens DPAF does not work 100% on my lens (but good enough, focus plane was always +/- 1cm right).

I'm not sure.  My Sigma 18-35 worked well on my 70D, and now it works well on my 80D as well.  In both cases I dialed in a little microfocus adjustment.  For both cameras, live view focusing works well - a bit more reliable than viewfinder focusing.

This is a lens I've been on the fence about for a long time. So many 80d users with big complaints. It's not often I hear anyone being happy with it but some are. Is it possible that there's this much copy variation? Seems unlikely. I don't like the idea of buying and returning several times.

jvceac wrote:

This is a lens I've been on the fence about for a long time. So many 80d users with big complaints. It's not often I hear anyone being happy with it but some are. Is it possible that there's this much copy variation? Seems unlikely. I don't like the idea of buying and returning several times.

It is my favorite lens by far.  I can't explain why others have such difficulty.  My only theory is that the lens is so sharp, and the focal plane can be so small, that any missed focus will be very noticeable.  Also, people tend to use the lens in fairly dark indoor environments, the most challenging of all places.  In comparison, my 50mm STM 1.8, is pretty sharp wide open, but not as sharp.  It definitely has a narrow depth of field.  But I usually use it outdoors, in much better light than my Sigma.

It could also be that particular camera firmware versions don't play well with particular lens firmware versions.  Seems to me that Canon 80D is Sigma's primary market, so I'd think they'd figure out the issue and fix it.

A few samples attached - 6 of 6 that I took.  Focus point was the center of the photo, at the base of the closest flower.  Half the shots were through the viewfinder, half with live view.  To be honest, I forget which is which.  Full disclosure:  in low light the focus reliability would be lower, quite possible that not all 6 would be close to perfect.

Thanks for that. Someone else, several months ago made similar comments regarding low light and thin DOF leading some to believe there were focus problems. Did you do any focus adjustments either in camera or with the dock?

jvceac wrote:

Thanks for that. Someone else, several months ago made similar comments regarding low light and thin DOF leading some to believe there were focus problems. Did you do any focus adjustments either in camera or with the dock?

Yes.  On both my 70d and 80d I made adjustments for each wide aperture lens I've owned, Canon and Sigma:

Sigma 18-35 & 50-150,   Canon 50mm II and 50mm STM

I never used the Sigma dock.  I never felt that the narrow aperture lenses that I own needed adjustment.  Hmm... I may experiment and pixel peep a little, maybe even those could benefit from a little adjustment.