Review: Sigma 50mm f/1.4 DG HSM Art Lens

[brad-man]

I'll wait for Canon to release an awesome and cheap 50mm 1.8 IS lens. I have a hard time dropping money on a 50mm lens for some odd reason.

Cheap?

[LetTheRightLensIn]

"occasional AF inconsistency"

Noooooooooooooooooooooo!!  This is why I retuned all of the copies of their old 50mm...  Hopefully the Sigma dock can fix this.

Otherwise the review looks great.  My biggest concern with this lens was if it sacrificed a nice smooth OOF for sharpness.  The few "Bokeh" (I never know if I am using that word correctly) pictures I have seen look good though.

Man what is it with 50mms?? Why can they never AF? Zeiss 50mm new or old or older don't AF. Canon 50mm 1.4 and 1.8 don't AF (1.2L maybe does if you escape focus shift). Old sigma doesn't new sigma doesn't. Zuiko doesn't AF.

[Invertalon]

I just tried my 4th... Yes, 4th copy of the Sigma 35... I have written Sigma off at this point. I tried, but I just could not deal with the AF.

The first two I tried right when the lens came out... MA changed heavily indoors vs. outdoors... 0 MA near to far outdoors, +10 required indoors. Was a PITA.

Fast forward to last week, I tried another. Copy 1 seemed to work fine indoors vs. outdoors but was decentered... OK, just get another... I have had plenty of Canon lenses that I needed to exchange.

Got the replacement, indoors to outdoors is perfectly fine (thankfully). However, the AF is just way too darn unreliable. Some shots are so far OOF you wonder how the camera ever thought it was right. Yet the next photo be perfectly sharp, the next somewhat soft, sharp again, way out of focus the next, etc... Mostly near infinity focus it had these issues.

I just slapped my 35L back on and sent the Sigma back home for the last time... Until Sigma can have even better AF, the Canon options are for me.

[skullyspice]

sounds like a dud. i'll stick with the 1.2

[Snodge]

A review before the lens is released? I think instead it should be a preview, not a review! I'd also like to see some other full reviews just in case there was something wrong with that particular lens that was (p)reviewed...

[Invertalon]

With that much trouble, one has to wonder whether the problems are with the lenses or somewhere else ... like the camera or the "driver".

Trust me... It is Sigma. When all of my Canon glass works perfectly in terms of AF and all (4) sigma lenses are all over the place, I think I can rule out operator error.

Optically aside from the one decentered one, they were amazing. The AF though is a whole nother story... Also note, two copies of the Sigma was on one 5D3 and the other two on another. So I can rule out the body as well.

The 35L nails focus. When it misses, it may be slight but usable. The Sigma misses by a mile at times. Makes no sense. With my commercial work I can't risk that just for sharpness. AF is very important. It's not like I didn't try though.... 4 copies over a year or more? It's not like I gave up after one...

[Dreamer]

I have to admit I have been waiting for the outcomes of the Sigma reviews following the success of the S35/1.4A.  I'm actually not sure why either - i have been reasonably happy with my C50/1.4 and I cant yet justify the $ for the C50/1.2L.  I say reasonably, but i mean really , as at first when i started using it was a bit hit and miss (focus, bokeh et cetera) and after hovering around here for a while realised this is what you have to accept with a relatively old, fast lens. 

Then I read about someone having super success using FoCal and decided to give it a go.  Having tried to micro adjust it a few times, I kept doing it until I got 3 successive scores that were the same (-3).  Well - it's been a revelation!  I love my C50/1.4.  For the price and rewards I get from it - I will just wait to see what Canon may have in terms of an upgrade in the near term.  That way i can stay within the same eco system with minimal risk and fuss.  Who knows - may lash out for the 50/1.2L, but don't think I deserve that just yet.

Just my 2 cents worth .  For me, in order to move away from Canon eco system, the rewards must really match the risks of investment (ie; price, quality, consistency, durability, resale).

[pwp]

"occasional AF inconsistency"
Noooooooooooooooooooooo!!  This is why I returned all of the copies of their old 50mm...

Arrrgh! Me too. The old Sigma 50mm f/1.4 was a beauty...when it focused. I went through two copies, trips to Sigma and finally gave up; thanks eBay.

-pw

[callmeasyoulike]

Is there a lens out there that never has occasional AF inconsistency?

Does a 40% miss rate really constitute occasional inconsistency?  I think not...

Compare this to Zeiss missing 100% of AF shots 

Well you could turn the AF off and then the Sigma lens would have the same rate of AF misses as the Zeiss lens, for $3000 less.

+1 

[TM]

Great review, as usual. Now I'm considering cancelling my pre-order. Was hoping it would be as solid as my Sigma 35mm f1.4 Art lens, but if that's not the case, I'll stick with that and my other Canon L lenses.

[Eldar]

Below I share ten 100% crops from one of the more-formal focus tests I performed. The subject is a large book properly aligned with the camera at a relatively close focus distance. Starting with a slightly defocused lens, each shot was autofocused using the center AF point that was very comfortably and completely covered by the book. The first 5 and last 5 images from this particular test are presented below and are representative of the larger test group. … The camera was a tripod-mounted EOS 5D Mark III with mirror lockup and the 2-sec self-timer in use.

Of those 10 shots, 4 are sufficiently OOF as to be unusable (3, 4, 6, 10).  A 60% hit rate with a static subject and a tripod-mounted camera, particularly one with an excellent AF system, does not inspire confidence. 


Also, this is a departure from the norm for Bryan's lens tests (and one, frankly, with which I'm not too pleased):

My evaluation lens was a short term loan from Sigma, as they offered the production-grade lens before it was commercially available.

Any time a manufacturer supplies a product to a well-known reviewer, a big unanswered question is whether the provided copy is truly representative of units purchased retail.  Clearly, it would be in Sigma's best interest to pre-test a batch of them and pick the best copy they can find for review (in fact, they are supposed to generate measured MTFs for every lens they produce, so they have the data already).

I've always felt that one of the strengths of Bryan's reviews (in addition to their thoroughness and readability) is that he purchases review copies through standard retail channels (B&H may put him near the top of the preorder queue, but that's fine), and therefore avoids the potential confound of bias introduced by testing a 'hand-picked' lens from the manufacturer.  I hope Bryan chooses to test one or more copies of the lens purchased retail to see if the results align with the copy provided by Sigma.

I totally agree with your views. I was a bit surprised that Bryan published this as a full review and not just a preview. His objectivity takes a hit when he writes this based on a copy provided by Sigma. On the other hand, he still provides what appears to be a frank, open and honest review. I hope he goes back and provides a revision when he get copies through his regular retail channels. I also would have liked to see more image examples.

I have the 50 Art on preorder and I am looking forward to see what it is worth. But my experience with the 35 Art, where the AF perfomance is a bit unpredictable, is that I end up leaving the lens in the bag (if a 35/1.4L II came around, I would definitely try that). If the same thing happens with the 50 Art, and especially because my main reason for getting it is to use it from f1.4-2.8, I'll probablly reach for the Otus instead.

[Rudeofus]

I really wonder what is so damn hard about reverse engineering that Canon lens protocol. We are not talking about some one man show running a startup on a shoe string budget and Ramen noodles, AFAIK Sigma is a sizable company that can design outstanding lenses. It's not like they'd have to crack AES encryption to make this work. They build up all this reputation for the new 50A, only to see it shredded by their poor electronics/firmware. 99-yard football seems to be their favorite sport ...

To those who wondered why AI works for the 50A and single shot AF doesn't (reliably): single shot AF is usually an one effort procedure: measure point spread, calculate AF motor movement, perform motor movement, done. If the measurement is off, or the motor does not move as intended, your AF will be off. With AI the measure/calculate/move procedure is performed continuously, and therefore will only fail in focus shift situations (see 50L).

[Albi86]

Besides some cases of users having problems with the 35A, I believe in the majority of cases the lens works quite well.

I can't imagine why this 50mm should be a step back. It is either a lens design problem (big issue) or a small software quirk that will soon be corrected via firmware update (small issue).

It's a pity because the rendering of this lens looks really stunning.

[candyman]

How many (p)reviews of the Sigma 50mm f/1.4 ART actually report problems with the AF? I read a few but only found this report at TDP. Is it possible that Bryan had bad luck with his copy? Although it was provided by Sigma?

EDIT: Lenstip reports an AF problem specific with the Canon 1Ds MKIII. Maybe they should have tried another FF to see if it is that camera or the lens....

I used to own the 'old' Sigma 50mm f/1.4. I never had AF problems with that one. I got rid of it because I did not like to performance on the FF but loved it on my 7D.
It is possible that production copies will perform outstanding where always you have some copies that are not okay.

[Viggo]

I really wonder what is so damn hard about reverse engineering that Canon lens protocol. We are not talking about some one man show running a startup on a shoe string budget and Ramen noodles, AFAIK Sigma is a sizable company that can design outstanding lenses. It's not like they'd have to crack AES encryption to make this work. They build up all this reputation for the new 50A, only to see it shredded by their poor electronics/firmware. 99-yard football seems to be their favorite sport ...

To those who wondered why AI works for the 50A and single shot AF doesn't (reliably): single shot AF is usually an one effort procedure: measure point spread, calculate AF motor movement, perform motor movement, done. If the measurement is off, or the motor does not move as intended, your AF will be off. With AI the measure/calculate/move procedure is performed continuously, and therefore will only fail in focus shift situations (see 50L).

It's not about the lack of skills, it's matter of not violating patents Canon own.