Need the speed? Sigma 18-35mm F1.8 DC HSM in-depth review

Yeah, just get some ideas from the Professional like you, so any suggestion for all around lens to my 6D except 24-70mm of canon & tamron they are costly.. what about the kit 24-105 but it was an old model. thanks for your reply Andy, hope you can help me to have a usable lens which is not soooo expensive : )

Went out shooting like any other day with my 70D and the auto focus missed too often. I never shot below 1/125 sec. The colour and contrast was very nice and when it hit, the images were beautiful. What's the point of having f/1.8 if the hit rate for keepers is so low. I returned mine as well. After reading this review, I was reluctant to try another copy. Too bad, I really wanted to like this lens.

No, it's not only a body issue. AF uses camera-lens comm and lots of calculations. I have tested carefully the Sigma 17-50mm f/2.8 and compared it to the Nikkor 35mm f/2, and the Nikkor performs much better re AF consistency and accuracy compared to the Sigma.

the difference is just under 5%.
the limit that we tolerate (or we used to tolerate?)

@KerryBE: 'We' in this statement refers to this very website, dpreview.com: I write all the lens reviews here. Here's a comparison between the Sigma 17-70mm F2.8-4 DC OS Macro HSM |C on Canon and Nikon cameras. The slight difference between sensor sizes has little impact on the lens measurements.

This is crap. Who can define a standard zoom? For that, we need first to fix that standard. If you look at other manufactureres, a fast Zoom with such an aperture is always short in range, the Nikon 17-35/2.8 for example, the Tokina 11-16/2.8 is another one. The faster a zoom gets, the shorter the range. Doing a zoom of a displacement of more than 20 mm with such an aperture is a challenge, and imagine the size you need to do this.

So, if Sigma brings us a 18-35 with 1.8, one can only say, "hats off" it's a performance. Now, if they did a 18-55/1.8, it would be the size of 300/2.8 barrel, and I doubt that anyone would buy it. My only concern is that it is somewhat expensive, but, i will not buy it anyway, I prefer to invest that money in a huge prime with 1.8 or 1.4 aperture.

18-35/1.8 is a standard f/2.8 equiv. zoom for APS-C SLR. Sigma did a great job that it's much better than small aperture 17-55/2.8 zooms from Nikon and Canon.

what's great is not the aperture, but a high quality wide-standard angle zoom for APS-S SLR which is the most difficult mount (that happened to become popular for some technical reasons we had more than a decade ago).

Thanks Andy. I read the review but the conclusion is incorrect. That nature of a PDAF system limits the utility of a lens-based MA feature. It's actually much more effective to adjust in the body than in the lens. A PDAF acquisition involves multiple phase-detect sensing and lens movement cycles. Altering the magnitude of any single movement by the lens (lens-based MA) is an out-of-band correction for which the PDAF system will readjust for in the next sensing/movement cycle. In other words, it's an alteration that is lost in the feedback loop.

Andy, the determination on when the PDAF system has achieved sufficient focus (within DOF marign) is based on the phase detection. The need for AF adjust occurs when the optical path through which phase detection is measured has a misalignment or lens aberration which affects the accuracy of the phase comparison. Focus-element positioning errors (mechanical errors) are self correcting via the multiple PD cycles. But those errors are detected/calibrated against the phase sensing mechanism, which has no self-calibration mechanism and thus requires AF adjust. It may seem like an armchair theory but it's based on my development of DotTune, which you can read about here: http://www.dpreview.com/forums/post/50774257 and here: http://www.dpreview.com/forums/post/50883023

And again - Sigma's in-lens microadjustment demonstrably works in practice, and that really is all that matters.

Our discussion is whether in-body AF adjust works better than in-lens, and my assertion remains that in-body is better. I've experimented with Sigma's in-lens AF adjust on the 35mm Art (compared with in-body AF adjust) and my empirical results match the theoretical assertions. You say the Sigma's AF adjust works in practice yet you still report that its AF is inconsistent. Perhaps you're defining "work" as an improvement over the baseline, or perhaps as just observing a noticeable change in AF behavior. Going back to my original post I still believe the 70D's dual-AF adjust for zooms has a good chance of yielding more consistent AF results than what you observed using Sigma's adjustments.

The Sigma's AF is inconsistent under less-than-ideal conditions. However, with a well-lit high-contrast target of the type used for measuring AF accuracy and determining microadjustments, it's much more consistent. Under these conditions, without in-lens AF microadjustments it's consistently (slightly) inaccurate, but with them it's consistently accurate. Therefore, in-lens AFMA works.

However, even after microadjust, the 18-35mm's AF remains inconsistent under less-ideal conditions on every camera we tested it on.

Thanks Andy. For reference, an AF adjust value that appears to be correct/accurate but is not can manifest as inconsistent shot-to-shot results, esp in marginal AF conditions. Can't say that's what's causing your issue but just wanted to mention it as a possibility. I have a post describing why here: http://www.dpreview.com/forums/post/50781856.

Also, was the lighting temperature for your low-light AF test the same as your abundant light tests?

shameful that Canon cameras are resolution challenged but D7100 got 14.8MP within 4/3" range (it's 13.4MP for 70D).

it's much better to be able to do post-cropping for we can have better framing and less chance to clip-out wanted image when composing too tight or shooting fast moving subjects.

For this lens even 3m is like miles away if you consider that the eye of the horse is so small and no contrast at all. Might as well focus on the white wall. Plus camera like 700D could easily be fooled to focus on the background with a lot of contrast. So if you have more misfocused examples but with better target we sure would love to see them. And also would love to see Canon 35mm F1.4 in this situation side by side.

I'm not sure. The Olympus 35-100 F/2 costs around $2500 and weights a little over 1.5kg.

@kimchiflower
The Olympus 35-100 F/2 was designed for the 4/3format, so it covers an area much smaller than an scaled 18-30 for APS-C. The Olympus designers could use relatively small rear optical elements, mitigating somewhat the effect of the "third- power law". Consider also that if the Olympus lens aperture were F/1.8, instead of F/2, the weight and cost of the lens would be considerably higher. To get an idea of how the weight increases fast with the aperture, compare a 50/1.4 lens with a 50/1.2.

@EKB
The "third power law" is a general trend, not an absolute law. If all the linear dimensions of a lens are scaled up by a factor k, the focal length increases by k, the f-number keeps constant, but the volume (and the mass) of the lens increases by k^3. In practice, this "law" is not so horrendous because the masses of the several parts are scaled only partially, or not at all, by the third power, as is the case of the mount. In general, the weight of the front optical elements tends to follow the third power law, but the diameters of the rear elements are related more to the sensor size than to the focal length, so the third power law applies only partially to the rear elements.

@Frank_BR
In the 18mm to 70mm range, the general trend is so weak as to hardly be worth calling a trend at all. In fact, can you even name a single pair of Nikon F lenses of 70mm or shorter focal range that fit this third-power trend?

If I had to guess, I'd guess that a 35-70mm f/1.8 APS-C lens would be somewhat heavier and more expensive than the 18-35mm f/1.8. But not by a factor of 8. In fact I'd be suprised if it were more expensive by a factor of more than 2. On the other hand, I would *not* be suprised if a 35-70mm f/1.8 turned out not to weigh or cost *any* more than the 18-36mm f/1.8.

Prime lenses for the FF format with focal length below 70mm can have very dissimilar design: retrofocus (35mm or lower), double-gauss (around 50mm) and telephoto (70mm or higher). Of course, direct scaling does not apply for different types of design.

You should compare lenses of similar designs, like the Nikkors 105/2.0 and the 200/2.0, for example. These lenses weight 640g and 2930g, respectively, with a ratio of 4.57. This value is below the "expected" 6.91 (1.9^3), but the discrepancy shouldn't be a big surprise because the weight of certain elements that made up the lens (mount, barrels, motors, etc.) does not follow the "third power law".

"Prime lenses for the FF format with focal length below 70mm can have very dissimilar design."

Yes. Exactly so. And the same applies to zoom lenses and to lenses designed for the APS-C format that have focal lengths below 70mm.

"You should compare lenses of similar designs"

Why? Why should I expect a 35-70 f/1.8 to have a similar design to the 18-35 f/1.8?

In fact, I would expect it to have a different design - to *need* to have a different design - in much the same way that, e.g., the Nikon 24-70 f/2.8 has a different design than the 14-24 f/2.8

As it turns out, the 24-70 has about the same cost and weight as the 14-24 - in fact, it's a bit lighter (900 vs 1000g) and a bit less expensive. By analogy, I'd expect a 35-70 f/1.8 to have about the same cost and weight as the 18-35 f/1.8 - not eight times the cost and weight.

"I'd expect a 35-70 f/1.8 to have about the same cost and weight as the 18-35 f/1.8 - not eight times the cost and weight".

I wouldn't be so optimistic. Consider first the Nikkor 24-70 F/2.8 lens that weighs 900g. Now consider a hypothetical 24-70 F/1.8 lens. The front elements of this lens would have a 2.5x larger frontal area compared to the 24-70 F/2.8. Usually, an F/1.8 lens has a much more complex design (more elements) and is physically longer than an F/2.8. What would be the weight of the hypothetical lens? Possibly more than 3kg, I guess.

In summary, with due respect to you opinion, I doubt it would be possible to design a 35-70 F/1.8 with the same weight of the Sigma 18-35 F/2.8.

1. I'm not considering a hypothetical 24-70 f/1.8 lens, I'm considering a hypothetical 35-70 f/1.8 lens - and one that APS-C rather than FF, to boot.

2. By your logic here, the Sigma 18-35 f/1.8 couldn't possibly exist either:

Consider the Nikon 17-35 f/2.8 that weights 745g. Now consider a hypothetical 18-35mm f/1.8. The front elements of this lens would have a 2.5x larger frontal area compared to the 17-35 f/2.8. Usually, an f/1.8 lens has a much more complex design (more elements) and is physically longer than an f/2.8. What would be the weight of the hypothetical lens? Possibly more than 3kg, according to this logic.

3. Or to put it this way: The old Nikon 20-35 f/2.8 weighed 588g. The old Nikon 35-70 f/2.8 weighed 664g. The Sigma 18-35 f/1.8 weighs 811g. So a Sigma 35-70 f/1.8 should weigh what? Maybe 900-950g? Or maybe less, because the Nikon lacks the weight-adding 2mm at the wide end.

"1. I'm not considering a hypothetical 24-70 f/1.8 lens, I'm considering a hypothetical 35-70 f/1.8 lens - and one that APS-C rather than FF, to boot."
-------------------------------------------------------

A hypothetical 35-70 F/1.8 lens for APS-C wouldn't be much lighter than a 24-70 F/1.8 lens, because in this case the overall weight would be given by the upper limit of the focal length, which is 70mm for both lenses.

"2. By your logic here, the Sigma 18-35 f/1.8 couldn't possibly exist either:
Consider the Nikon 17-35 f/2.8 that weighs 745g. Now consider a hypothetical 18-35mm f/1.8. The front elements of this lens would have a 2.5x larger frontal area compared to the 17-35 f/2.8. Usually, an f/1.8 lens has a much more complex design (more elements) and is physically longer than an f/2.8. What would be the weight of the hypothetical lens? Possibly more than 3kg, according to this logic. "
--------------------------------------------------------------------
to be continued…

That was not a proper comparison! Even though those lenses have almost the same focal length, their angular coverages are very different for the wide angle end: 104º versus 73º. The difference of coverage makes a profound impact on the lens design. The Nikon is an ultra-wide- angle retrofocus zoom lens of very complex design, whereas the Sigma works in the relatively easy range of 73º to 43º. To have an idea of how difficult is to design a lens for a 104º angular coverage, please take a look at the Fig.29 (*) at:

https://dl.dropboxusercontent.com/u/60804047/British%20Journal%20of%20Photography%20-%20Feb.%201980.tif

Sadly the graph stops at 90º, but it is reasonably to conclude that a 104º angular coverage (Nikon) would demand 3 to 4 times more optical resources than 73º (Sigma).

(*) excerpt from "New Developments and Trends in Photography Optics at Zeiss", by Walter Wöltche, Head of Mathematics Department, Photo Optics Division, Carl Zeiss, Oberkochen

to be continued…

"3. Or to put it this way: The old Nikon 20-35 f/2.8 weighed 588g. The old Nikon 35-70 f/2.8 weighed 664g. The Sigma 18-35 f/1.8 weighs 811g. So a Sigma 35-70 f/1.8 should weigh what? Maybe 900-950g? Or maybe less, because the Nikon lacks the weight-adding 2mm at the wide end."
-------------------------------------------------------------------------

The Nikon 20-35 F/2.8 lens is an ultra-wide-angle, so it shouldn't be used in the comparison with the Sigma. However, I think it is valid to compare the Nikon 35-70 F/2.8 with a hypothetical Sigma 35-70 F/1.8. The question is: How heavy the hypothetical 35-70 F/1.8 would be? I don't know for sure, but, again, the Fig. 29 again gives a hint. Although the graph stops at F/2.0 and is specific for a normal lens, the Fig. 29 suggests that the cost of raw optical glass for an F/1.8 lens would be several times more than for an F/2.8 lens. This agrees with my perception that the weight of a 35-70 F/1.8 lens would be several kg.

Glad to hear I'm not the only one completely sick and tired of this equivalency nonsense. I use APS-C and full-frame 35mm digital. And (real full-frame) medium format as well. I even dare to use film sometimes.

I am odd, perhaps, but I do not walk around all day making equivalency computations and DOF charts, "light gathering" comparisons and whatnot.

I shoot.