Sigma 20/1.8 vs. Canon EF 20-35/2.8-Battle of the Super Wides

What if I want to compare the Sigma 20/1.8 to the Canon 20/2.8? Or perhaps the new 17-35 at 20mm? There's no way I can take a photograph of the same scene again, and I certainly don't want to take, expose, and scan 6 photographs to make those comparisons. This is why real lens reviewers use USAF resolution charts and create MTF curves.

What is the purpose of the graphs in the "Personal Assessment" section?

IMHO, a more useful comparison would be the Sigma prime and the Canon prime or the more recent zooms (17-35 or 16-35) since these lenses are surely more common than the 20-35.

I am still trying to figure out the point of comparing these two lenses. There are too many variables. First of all, the Canon 20-35 has long since been discontinued. Sure you can find it on eBay. However, if I go as far to look for it in the used market, I would have a totally different motivation--perhaps because the lens meets my focal range need, is a Canon, is an "L" lens, and the used price is right. I can't see how the lens compares to the Sigma 20/1.8 is relevant at that point. If your choice of lenses was to level the playing field by price, the fair comparison should have been against Canon's 20mm f/2.8. If you must compare the discontinued Canon 20-35mm f/2.8L, you should compare it against the Sigma 20-40mm F2.8 EX Aspherical DG DF, which goes for about $600 brand new.

I do have an objection to this review, but it has nothing to do with the concept itself. It's with the comment that one needs to re-polarize with the EF 20-35 2.8L. Well, I bought mine new about 10 years ago, and it's internal focusing, I've never had to repolarize. Perhaps I misunderstood, I only skimmed the article, but just to make sure I wasn't completly daft I ran downstairs, picked up the lens, turned it to manual focus and focused it, the lens never changed length at all, nor could I see the front element changing length. There obviously was an element close to the front moving in the lens, but the very front one was stationary. Anyway, perhaps there were more then one versions of the 20-35 2.8L, or perhaps I misunderstood why the author feels the need to adjust a polarizer after focusing.

Evan

I still fail to see the point of the comparison, but you have a good review of these two lenses, it just happens to be written in the same article.

Why use BMPs that aren't supported in many browsers and are very large, when PNGs are supported in most browsers and have pretty good lossless compression (unlike JPEG's lossy compression)?

PNGs even support things like 16 bits per channels (48 bits per pixel) and embedding ICC profiles, though I doubt browsers supports the former or use the latter (but if you're interested, use a separate viewer).

Very nice article. To some (myself included), photographic examples of lens differences make a bigger impression than USAF, MTF, etc. data. Numbers are great, I'm an engineer for goodness sake, but everything needs a frame of reference. For those that get hung up on the massive differences in the subject lenses, for you maybe the subtext of the article should be "Does the prime-is-always-better thought apply to wide-angle lenses?"

For those of you reading this to evaluate the Canon 20-35mm please pay special attention to the head to head shots. Note that the foreground is in focus in the Sigma shot and the background is in focus in the Canon. The f/2.8 comparison is ridiculous. I own a 20-35/2.8 and I would like to say that it does in fact live up to it's reputation. Sharpness, distortion control, flare and ghosting control, chromatic aberation, vignetting, and color rendition are all top notch when compared to other zooms and primes that I have used. I use it wide open often and am very happy with the results. It is the only wide angle zoom I have used which requires little to no correction for light falloff and distortion when stitching panoramas together.Cheers

I thoroughly enjoyed your review. All the critics sometimes forget that amateurs and enthusiasts still peruse through the site and real-world examples are much more helpful than charts and graphs. We're not all pros you know ;)

Matt is correct. The choice of a non-planar subject completely invalidates this test, especially at wide-open apertures. The lack of "sharpness" we are seeing is really a lack of DOF.

While the painstaking work in this comparison is commendable, the showdown many photographers would like to see right now is between two new zooms: the Canon EF 16-35mm f/2.8L and the Sigma AF 15-30mm f/3.5-4.5 EX DG Aspherical.

Comparing these zooms is especially relevant for anyone with a digital SLR in their present or future, since most of their current wide lenses will become "normal" focal lengths on the digital body when the 1.3X or 1.6X focal length multiplier kicks in.

With the Canon costing about three times the Sigma, the latter is a bargain if it even comes close to the quality of the pro zoom.

Has anyone seem reputable reviews of either of these lenses on the web or in print?

The histograms are from scans? I assume you attempted to maintain strict manual scanning settings that are consistent for all images. I didn't understand your inclusion of histograms when reading the article and I'm not sold on your explanation in the comments, especially knowing that the results are dependent on your scanner, software, settings and workflow. I think you make your basic points without needing to add the confusing and distracting histogram information.

For what it's worth, the Canon 20-35L is quite good at the long end, sharp even wide open, though it becomes less good out to 20mm. I used it successfully for several years, built like a tank too.

This review answers a few questions that many people (Myself included) have had regarding "pro" zooms vs "consumer" primes - the most basic of these - Are they worth the extra money?

(This review seems to indiacate that the pro zooms in general are not.)

The histogram information may or may not be important to anyone/everyone but it provides the reader with a baseline relevant to the review and the reviewers equipment/software used in the review.

As for me , the review was informative and appreciated - and I dont use Canon or Sigma lenses.

When people start looking to add a lens to their collection, it is very common to start pondering: shop for current or older model? shop for a used or new lens? Are older optics good enough when compared to older ones?

Consequently, the review does have a place in the forum and does add value (more real life testing than MTF charts)..

However, comparing a larger set of lenses (maybe with some other brands nikon / pentax / tamron etc. , older / newer versions, prime / zoom lenses) and maybe more than one sample would have been great.. but also a full time job... Something is better than nothing.

Many thanks to Yaron for getting the ball rolling

Marios

I'd have to agree with Matt and Fabian's concern about the head to head f/2.8 shots. It really looks like the two lenses are simply focused at different distances. The Canon is clearly showing a much sharper background, even where the background is at the edge of the frame, while the Sigma renders a much sharper foreground. It gives the appearance that a depth of field difference is obscuring any difference in sharpness between the two lenses.

Yet the reviewer responds that both lenses were prefocused at the same distance. Maybe there's a difference in the two lenses' autofocus performance? Or if manually focused, maybe there's a difference in the focusing scale that caused them to be off? If so, that would be a more interesting and useful result than a difference in sharpness.

Well I for one found it very helpful. I am currently evaluating (much lower end) superwides for my new landscape photography hobby. My choice right now boils down to either a 19mm or 20mm prime or a zoom similar in range to the Canon. If nothing else, the discussion of light falloff and sweetspots was relevant to my choice.

Something to remember: those of us with SLR's (and not rangefinders with their lack of mirrors) will always have to make compromises with wide-angle lens geometries (no SLR based biogon-ish types available).

Considering that both lenses performed inadequately at their largest aperture, do you think that, perhaps, it is pressure from the marketing departments that get the companies to produce lenses that they can promote as an f1.8 even though it is unusable at that stop?

I would expect that this would not be limited to Canon and Sigma but that many of the specifications for lenses might be...optimistic...to say the least.

I wonder if the luxury brands...like Leica... are more accurate in their marketting and are usable "wide-open"?

Thanks for taking the time to write up your evalutations of these lenses.

I just wanted to add my thoughts on the usefulness of the Sigma 20mm f1.8 fully open.

No one with any sense is going to use this lens wide open in bright sunlight to shoot a landscape. You use it inside a smoky diner, a mud hut light by wood fire, or inside a theater where flash/tripod is not allowed or practical.

In these types of journalistic situations, critical sharpness is neither necessary nor is it desired. having the speed is the difference between getting the shot, or getting the blur from camera shake.

I have had this lens for six-months now, and can not tell you how many times I have looked through the finder and seen meter readings of 1/20 at f1.8. If I had the slow Canon zoom, I would not have the shot.

No, these pictures are not as sharp as those shot with my Planar 50 f1.4, but if you want the superwide perspective in low light without going to grainy faster films, this lens is really the only game in town. And it is a pretty good one, at that.

Unless of course, you can afford the Canon EF 24 f1.4L. But then, why would you be reading this????

Another nice feature of this lens is the close-focusing capabilities which allow you to do some neat stuff.

All-in-all, I have to say for $ 359., this lens is quite the bargain.

Now we will see how long it holds up in the field.....

Was either lens bench checked or calibrated prior to the test? Absent that, any comparison is a crap shoot and tells you only about the particular lens being tested, NOT other copies of that lens.

In my experience, used lenses may not meet original specs due to wear and tear that may not be visible.

This is also true of new lenses. One patron of a local camera store would buy three examples of a lens he wanted to acquire. He would test all of them, keep the best one and return the others. He paid top dollar for the confidence of knowing he had the best lens he could get. And he could document the differences between the examples.

FWIW/Scott Gardner

Both lenses were focused manually, to prefixed distance of 50cm, in order to obtain an hyperfocal-shot with the field of focus starting at 30cm (12 inches) to infinity, at f/2.8.

This just indicates you don't understand what "hyperfocal" means. It simply means that the shot is acceptably close to being in focus within that range. What is "acceptably close to being in focus"? That depends largely on the anticipated size of the resulting print. Most manufacturers make their hyperfocal marks based on the assumption that the resulting print will be small. No 35mm lens maker does so based on the presumption that you'll be enlarging to 20"x30", which you state is the equivalent of the magnification you're showing us.

Regardless, the fact is that those points closer or further away from the exact point of focus are going to be less sharp than those points at the precise point of focus. You're measuring a combination of bokeh and edge-to-center falloff in sharpness. As was mentioned above stopping down will bring elements in the frame not at the precise point of focus into sharper focus due to increased DOF. In this case you're measuring a combination of increased DOF and increased edge sharpness due to stopping down. In all cases there's no way to tease out the effect of the various factors on the test results.

In short, by chosing a non-planar subject you've built a meaningless test. The fact that you don't apparently understand this makes your entire review suspect. If I were still a photo.net moderator I would've rejected this review, frankly, on that basis alone.

You should reshoot with a planar (flat) subject and post those results. They would be both meaningful and interesting.

Light fall-off is another issue and this part of your testing scheme seems OK since the scenes are at least fairly evenly lit.

I might also add that people don't generally buy high-end zooms with the expectation that they'll be as sharp as a well-constructed prime. They buy them because they zoom. That's the whole point. If all you want is a 20mm lens then a 20mm prime is the clear choice. However, the 20-35/2.8L will outperform that 20mm prime at 35mm and all points in between. People buy high-quality zooms to replace a set of primes, not a single prime. Thus the very premise of this review is suspect. If all or nearly all of your use of a zoom will be at one end or the other of its range, then you're undoubtably better off with a decent prime.

Yaron at first I thought you did a very commendable job on your report. It is apparent that you did spend a considerable amount of time and effort on this endeavor. I was doing a search for a lens in this category. It has been several years since I have visited this site, however I am pleased to see that the Photo.Net Police are still alive and well. And that is the reason I do not frequent the site anymore. As Don Baccus points out you are a complete idiot and have no business posting such an inferior report and comparison. I sincerely hope that you will refrain any further exhaustive tests. Please do not waste our precious time.

It is insulting to the powers that be that you wannabe photographers have a question or conclusion that they themselves did not know the answer to when they began their quest to learn. They are the Hawks of Photo. Net. Although no longer an official monitor, the net.cops are always vigilant for dummies like you. They are the proud, perched, and ready, able to spot a novice at a single glance,

I always thought that the way to promote the art of photography was to encourage new folks, young or old, to exchange ideas, thoughts, and experience, in a friendly Non-Confrontational manner.

Questions and mistakes are how we learn. We need mentors not brow beaters.

Apparently the Net Cops do not agree, however I would only ask them to look back in time when it was new to them too.

And if you go spend money based on this so-called commendable review because no one has pointed out the flaws in the reviewer's testing methodology, you'll be happy after you find out that the methodology is crap? In this case it doesn't really matter because this is such an apples-and-oranges comparison, but in general sound methodology is a prerequisite for publishing a review of this sort.

When writing a review it is important to remember than some readers are likely to make purchasing decisions based on reading it. It is important to get facts straight for this reason alone.

How is the reviewer going to ever learn if the flaws in methodology are not pointed out? As I said in my first post, reshooting the tests with a planar subject (doesn't have to be a USAF test chart, old newspapers taped to a wall will work fine) the results would be both useful and interesting. I encourage him to do so.

Or the reviewer could leave out the meaningless resolution gibberish, and write a subjective piece explaining why, for him, a fixed, fast, 20mm lens is more useful than a wide-angle zoom.

That would be fine, too. I personally am not set up to run resolution test of professional caliber, which is why my reviews on this site are informal and subjective and make no pretense of being technical reviews.

re; the commentary... as an amateur Photographer but professional artist I spend many Saturday morning with photography buffs out Shooting. What amazes me is while many have an almost incomprehensible ability to state specs, and what to use and where... I almost always hear the same thing eventually. They buy a good quality auto and have fun. The reason why they seem to continue to come and the group grows is that what they really want is how to compose. I always seem to find the most technically proficient is usually the poorest composer. What to shot? not how is the final question. I have always loved photography and have always used a canon and zeiss but find I use the digital more and more,. The reason? I make notes on what colours I see, I know the colour wheel, I know what colour mix to make others.. . I now always encourage my group to make a detailed study on the sky the colours the greens etc. I have some who take a value finder to the shout.. I may have created another monster... BUT as one of these people said, HAVE SOME FUN. I make my living teaching painting and painting for the print market..I don't love it { prints} but people seem too. I photograph for me... and I love the freedom. cheers P p.s. I have a simga AND an canon.. go figure

Hi Don, James, and the others.

I had taken some time off to think over your comments, mainly because I felt this debate is heading towards a flame-war, something I really don't want to associate with my beloved hobby. I'm sure none of you think differently.

Don, I am the last person on Earth that would claim that he knows everything about photography. I've got a lot to learn, and quite frankly- I'm looking forward to it. And yes, I agree with you that I had diverted none of my resources towards investigating how these lenses deal with perspective correction; not because I chose to ignore it deliberately, but simply because It does not interest me much. True to your words, this may have given a (literally) distorted image of how these lenses perform. For this, I apologize.

For what does interests me -- lens sharpness and color rendition -- I believe that this article provides good real-world examples- something that is normally missing from other lens reviews. Perhaps I pay too much attention to sharpness and color, but that is how I use my gear. I assume that most readers here have capable skills of operating their gear to their own liking, thus they already know that nothing beats a first-hand experience; Besides that, reviews that are non-subjective tend to turn flat and uninteresting.

Aside for the testing-methodology, one thing needs to be remembered here. Both lenses were exposed to the same testing environment. Flawed or not, both lenses had to deal with the same composition, the same film, and the same exposure. Given these facts, the Sigma resolved more detail. Does it distort the image more than the Canon? I can’t answer that, because I hadn’t checked.

A zoom is a zoom, and a prime is a prime. Yes, they are different. The Sigma is NOT a zoom lens. It is inferior to the Canon at 24mm, 28mm or 35mm. If you need a zoom, buy a zoom. If you have money, buy both. Buy overlapping zooms. Buy every prime there is out there, and a sherpa to carry it for you :). Most of us don’t have this fantasy budget of course. I had to choose between a discontinued zoom and a prime; I was lucky to have both for a while and compare. I’m sure most none-professional photography hobbyists out there had the same dilemma, and this may had given them another point of view (Before, I was sure that L class glass would out perform any consumer lens, zoom or prime).

James, thank you for your support. I did invest my time and money into this review, and I believe a lot of people had profited from this article. You know how it is- most people are readers only, and I presume that the majority did appreciate this voluntary work.

I shall update the article to include more images. Unfortunately, I do not own the 20-35L anymore, and I don’t have images of a planar subject.

Cheers,

From the article: 'Do any of you still shiver out there when they hear the word "Sigma"?'

In a word, "yep". www.photodo.com having run out of funding (last real update in 2000), I turn to www.photozone.de for a semblance of objective test comparisons.

[Do check the methodology and disclaimers on the photozone.de page - the author is quite clear that your mileage may vary].

Still, the Sigma 20mm f/1.8 EX ranks just exactly LAST of all the lenses listed for optical quality. The CURRENT Canon 16-35mm L superwide zoom ranks far above it for optical quality, and even the creaking Canon 20-35mm L tests out far above it. As does the Sigma 15-30mm.

Check http://www.photozone.de/2Equipment/easytxt.htm for the complete list.

Not that I really care Pat, but on the same website: All Fixed Focal Superwides for Canon EOS, The Sigma 20/1.8EX ranks 2nd (above the Canon EF20/2.8, and the Canon EF 24/1.4L). Btw, can it be that one lens would rock on one mount and suck on another?

If we were all as emotional about auto windshield glass as about photo lens glass, road rage would kill tens of thousands a year.

A few points.

Regarding the validity of doing the comparison in the first place: if you never have to choose between buying a prime and buying a zoom, the comparison is not valid. If you do, it is valid. Criticizing someone else for making a decision like this on the basis that you don't make decisions like this is simply a waste of time.

Regarding "fairness" of the comparison: When you're trying to choose between two manufacturer's lenses with different characteristics, do you think "that's not fair! they should be the same!"? Or do you think something else?

Regarding Fabian's view that using a three-dimensional subject renders the test useless: this is true if what you intend to use the lens for is photographing two-dimensional objects.

If you intend to photograph three-dimensional objects like trees (I know, I know - a REAL photographer like a Galen Rowell or Jim Brandenburg would never do this), then you might be interested in a test methodology which includes a three-dimensional object in the frame.

Regarding depth-of-field versus depth-of-focus: given that the comparison is between images of the same object, at the same distance from the film plane, taken at the same focal length, I'd appreciate it if someone could explain why there SHOULD be a difference in EITHER subject sharpness or depth-of-field. The circles of confusion should be the same (both are on 35mm film enlarged to the same size), and the depth-of-field should be the same too.

Regarding Scott's comment that you have to calibrate and check lenses before you test: they don't calibrate or check them before they take them out of the glass case and put them in your shopping bag. If one lens is poorly calibrated, that suggests to me that either the manufacturer's quality control is poor, or the manufacturing tolerances themselves are very loose. Either way, to quote the review: AVOID.

Regarding the difference between testing lenses and using them: the smaller the difference the better.

Yaron,

Some observations about the photozone.de lens rating data you cited - which are also illustrative of the sorts of problems in all kinds of comparative analyses.

1) The table in question are "user ratings", and as such raise cognitive dissonance reduction issues ("I bought it, so I like it"). This effect is so well known, it defies argument.

2) As the AUTHOR of the table notes, the Sigma rating is in an "ORANGE font [which] indicates that the rating is stabilizing but still not overly reliable"

At that, it will be only a reliable report of how SIGMA users or CANON users of photozone.de feel about their lenses - not an objective comparison between lenses.

3) In any event, the numeric differences in the table are not statistically significant.

I'm still hoping someone will post the URL of an objective comparative lens database...but then, "reviews that are non-subjective tend to turn flat and uninteresting." Perhaps they are merely useful.

Photozone.de itself is obviously insufficient, unless the author divulges more of his sources and methodology.

Cheers. I shoot with a 20 mm prime myself - while other's are right that zooms are great because they zoom, they're also heavy. I have a weight set, and shoot static scenes - I don't really need the big ol' Nikon zoom as long as I can walk..

Pat, I agree with all of your comments.

Regarding my question- can the same lens design, of the same manufacturer, perform differently on different mounts? (Nikon/Canon/etc.?)

Cheers,

For another comparison of a prime with a zoom, may I direct you to this review over at luminous-landscape. Curiously enough, the methodology is much the same, but they also added a test for flare.

A quick google search found a couple of links that discuss MTF curves vs real world performance, this one at photo.net for instance. As several people pointed out, when testing lenses, your mileage may (will!) vary.

The comment made by a few regarding depth of field factors arguably invalidating corner sharpness analysis is certainly evident in the photos of head-to-head lens comparison at f2.8 with the images of foreground pinecones, center view of a single mossy tree, with trees and a little blue sky in the background. The review comment that the Sigma is “the clear winner in terms of sharpness” is belied by closer examination. The Canon lens is not only sharper at the center object in the image (the single mossy tree), it is more sharply focused all the way through the background. Plainly, the DOF of the Sigma is starting closer to the front of the image and fades out before the single tree, whereas the Canon’s DOF starts further out than the Sigma, missing clarity on the very close in cones, and kicks in as you reach the center tree and on through. The review didn’t say what focusing mode was used on the EOS 3 and the lens, but it’s possible that the focus points were not set the same between the two shots.

Thank you for comparing the two types of lenses available in the 20mm range (fixed and zoom). Most people will have to decide if the attributes of the fixed prime Sigma lens (lower cost, speed, weight, close focusing ability, sharpness, less distortion) outweigh the framing capabilities of a quality zoom like the Canon for their style of shooting. Ranting about minutia is fun but getting the shot is what it's all about. You have supplied us with more facts with which to make our decision.

Don

Although this review of two lenses is obviously frought with errors, I feel it was nicely done. I, for one, enjoyed it. It shows a lot of dedication and hard work to the HOBBY of photography. There were also some very interesting points made along the way. Points that should be very useful to a great many photographers. It was an honest attempt to test a pair of lens in the real world verses an old Air Force chart, etc. Would I use it as a buying guide? Probably not. However, it would cause me to take a long hard look at any used Canon 20-35 I might be interested in. It would CERTAINLY not cause me to jump on the Sigma bandwagon!!!!!!!!!!!!!!!! (I might add I have the 15-30 Sigma and 24-70 f2.8 and think they are marvelous, very sharp lens for my Fuji S-2!)

We learn by doing. Congradulations on your considerable efforts, amateur as they are. I am surprised at the rock throwers. Although Canon users are a dedicated bunch!

A point about the Pat Collins comment from Sept. 13.....

The 'cognitive dissonance reduction issues' defy argument? You said yourself that the table in question is composed of user ratings. So, these 'cognitive dissonance reduction issues' will apply to every rating of every lens on that table, not just the Sigma lens in question, because presumably every lens rated has been bought first. Your argument doesn't stand up because these 'issues' will be intrinsically incorporated into the rankings.

Yaron, nice review by the way. I don't agree with Don Baccus's earlier implication that a review should be perfect because someone could buy an expensive product based on it. Anyone who buys something as expensive as a lens based on one reviewer's opinion deserves everything they get if it turns out to be disappointing. Most people will look at a broad spectrum of reviews and opinion in magazines and sites like this, and then make a more informed decision. Amateur reviews like this one are useful because they just throw another opinion into the general pool of information available, upon which people make their purchase decisions.

"They also perform the same in terms of color (I still lean towards the Canon for its more saturated colors)."

You say they perform the same in terms of color then you say you lean towards the Canon. I don't get this comment.

I have not come across a Canon 20-35/2.8L with a rotating front element so far. I had a chance to play with one yesterday for a while, and no, the front element does not rotate, nor does the lens change length when zooming.

A nicely written and well-intended review, although not comprehensive. I suspect in one or two of the side by sides the lenses are focussed at different points.... Still, there is some useful information here, and some of the comments above seem a bit rude.

[no known relation to will chapman]

I just finished conducting a similar test, using both planar (resolution targets) and 3d (real photo) targets between my Nikon 17-35/2.8 and the Sigma 20mm 1.8 . While not recommended, I conducted the test using AF. At any aperture the Nikon had an edge at the center, but blew away the Sigma at the corners. Furthermore, the Sigma's corner performance was very inconsistent between the four corners, a characteristic apparently of a lens that is somewhat decentered. While three of the corners were generally in the acceptable range, the fourth corner (interestingly the upper right) was never acceptable for the planar target. I went to great lengths to ensure that the film plane was aligned with the image plane. The Nikon's corner performance was generally consistent between corners. I might add here that the corner resolution of the Nikon at 2.8 is better than the Sigma's at any aperture!

The only reason I bought the Sigma is the extra 1.3 stops of speed. This is even more true now after testing since at 2.8 the Nikon is definitely sharper with much better contrast. The only question now is whether the Sigma is sufficiently sharp overall (center and corners) for me to use wide open, especially on the planar targets for which it is intended. Frankly, even after shooting two test rolls, I still do not know the answer to that question, and will have to do some additional testing.

Incidently, if any are interested I have some lens testing instructions and targets at http://members.cox.net/lenstestr1/lenstest.htm

I have ( had ) a Sigma 28-70 DX EX series lense. When my camera fell off the coffee table it broke the lense off just past the aperature ring. It was mounted on my DCS 620 - (F-5). Very good pictures but this lense would not hold up in the field. The 20mm 1.8 EX is made the same. Good glass cheep housing. That my two cents worth.

Note: I have ordered a Nikon 17-35.

I'd like to add my thanks for your time in putting together this review. This forum and others like it only run on the energies and efforts of people such as yourself Yaron. Anyway... I've owned to 20-35 for many years and have some comments to add: My front element did not rotate, nor did the lens change size! I'm not aware of any revisions to the design so have to wonder what lens you where using? Most of my work has been photojournalism, ergo, I've used it wide open more than any other setting. It hasn't let me down once, there are some examples on my photo.net page. Oh yeah, it is built like a tank and I like tanks...

I would have liked to have seen the Sigma prime compared with an equivalent focal length Canon prime instead of the Canon zoom tested in this article. It would have been much more helpful for those interested in this fast Sigma lens. Alternatively, a Sigma zoom could have been compared to the Canon zoom if the main goal was to evaluate the Canon lens.

Just wanted to say after reading your lens review (20mm Sigma vs Canon...) that the comments were quite obviously Cannon users who were in a state of shock at the obvious results. Thanks for the review. It is important to me 1) cause I am thinking of buying the Sigma 20mm since I have a SD9 (digital camera by Sigma). This camera only accepts Sigma lenses cause of the size of the mount. I am looking for info about primes vs. zooms and your article was great. I can't even find another article that even test this 20mm Sigma lens. I have the 24-70 EX 2.8 and was hoping to find that the 20mm 1.8 was sharper. As a note the 24-70 is really nice, that is comming from a pro shooter. I bought the Sigma only cause its the only camera that has the superior Foveon chip housed inside. Chris

I've changed my mind... I've just returned the 20mm f1.8 Sigma, it's rubbish in the context that it's sold as with a "pro" designation. I put 15 rolls of numerous high-res slide film through it - mainly Veliva 50 and Provia F - while attached to a Canon EOS3 on a tripod. Images were scanned digitally on the ultra-new Minolta 5400 scanner :-) Lets cut to the chase, colours and tonal range are bland rubbish - obviously only relativly low grade injection-moulded optical glass has been used in it's construction. The reference/control point for the above statement is a comparrison with a few frames shot with a Canon 50mm f1.8II and EF 28 f2.8 on the same test rolls of film. There is significant edge blur at all apertures to the point that it struggles to resolve blades of grass and such like at the edge of the frame, but it manages reasonably well in the center. However, out of focus highlights are good. The sweet-spot just means it's less worse than wide open. If you are taking a picture of concrete, or some other subject with little colour variation, then all you're likely to get on the negative is a single tone with no detail - the same goes for a field of grass. I think you were lucky with your test shot, or you aren't scanning at a high enough resolution or using real Pro film. Quite how you can choose the Sigma over a Canon L zoom beats me. No offence, but you must be subconsciously delusional and therefore biased toward portraying the Sigma to be an “ok” lens in order to justify saving money. Do you work for Sigma, are you a real user; are you deliberately misleading people? The Sigma is ok for 4-5megapixel digital work, but not even that if you in the slightest bit critical and observant. Sigma are still "trying" in my opinion and the build and quality is akin to cheep "Jap" cars; their 170-500 lens is the only Sigma lens I would consider using. I'm quite offended Sigma have wasted my time (yet again!!) with another over-hyped lens. If they positioned it as a mass-produced regular digital-only prime then fair enough, I wouldn't have given it a second glance, but... I'm saving up for the Canon 20mm EF which wipes the floor with the Sigma, as does the EF 17-40L.. Thanks, Al.

concerning the choice of test shots: is it not the case that the cheaper lenses are likely to show their limitations shooting towards the lightsource? esp. strong sunlight. Since they use inferior coatings? That was my experience with sigma 24mm f2.8. though it scores highly if you measure the MTF, it performs horribly shooting towards the sun, even with the lens shaded and sun excluded from frame, the worst aspect being colour rendition. (my contax wide-angle doesn't suffer this problem to nearly the same extent). Whilst shooting away from the lightsource it performs pretty well, especially stopped down. So comparisons between "pro-" and "consumer-" grade lenses should include a shot towards the sun; I guess this is partly what you are paying for with "pro-grade" lenses.

Hmmm.... interesting array of comments and pro/con Canon (or Sigma), the usefulness (or non usefulness) of MTF charts, et al, but....

... not one comment on AF speed!

Goes to show the wide variety of criteria by which photographers assess the usefulness of a particular lens in comparison to other lenses!

Also note that I have never had a client assess my images with reference to an MTF chart or in direct comparison to another lens. ("Oh, you should've used a 20mm f1.8 Sigma, not your L Series 24mm f1.4 Canon!"

.... and I'm still wondering about the AF speed of either lens.