Why the EFS 10-22 is likely an optical gem

[Jonathan_Kwok]

Hi everyone... it's my first time posting here and thought I'd share some thoughts and speculations on the performance of the new EF-S 10-22mm USM lens, seeing that there's been considerable discussion about this long awaited lens.

Let me say right off that I’m highly optimistic this lens will yield outstanding performance although the first EF-S lens was just mediocre at best. Based on several factors, I’m willing to bet that the EF-S 10-22 lens will equal traditional L lens quality.

Here’s why.

It’s a fact that wide-angle SLR zooms are the most difficult lenses to design and as far as Canon is concerned, their wide-angle zooms have traditionally been the weakest link in the chain of EF lenses. However, they’ve been working hard to improve their wide-angle zooms. Consider the evolution of wide-angle lens design in Canon. The company has released five versions of superwide zooms since the Eos series started - EF20-35 f/2.8L, EF20-35 f/3.5-4.5 USM, EF17-35 f/2.8L, 16-35 f/2.8L to the current EF17-40 f/4L. With each successive generation, the optical performance has steadily improved and the designers have seemed to arrived at what they consider an ideal optical formula based around the EF16-35L’s design.

The EF16-35L lens has a 14 element, 10 group optical formula utilising 3 aspherical and 2 UD lenses. The newer EF17-40L lens has a 12 element, 9 group lens design utilising 3 aspherical elements and 1 super UD element. The 17-40L is obviously a refinement of the 16-35L’s optical formula with one Super UD element replacing two UD elements. The use of fewer glass elements and smaller max aperture in the 17-40L means that it is generally less susceptible to flare compared to the 16-35L.

Designing a new optical formula takes too much time and money and I’m sure Canon designers did not take this path for the new EF-S lens. They have merely modified existing lens designs from the EF series to adapt them for EF-S mount use. Ergo, the EF17-40L is reincarnated into the EF-S 10-22, while the EF28-135 IS USM becomes the EF-S17-85 IS USM. So, from a design standpoint, the indications are good that the new EF-S lenses should deliver the goods, especially for the EF-S10-22mm.

Now, with this in the background, consider the optical layout of the new EF-S10-22 USM. It has a 13 element design, utilising 3 aspherical and 1 Super UD elements. A check of its layout immediately reveals that its optical formula is based on the tried and true EF17-40L design, which is arguably Canon’s best wide-angle zoom currently.

The use of exotic glass and the familiar optical layout should put the EF10-22 in the same league as L lenses. I’m guessing that for market positioning reasons, Canon has chosen not to give it a prestigious thin red line. Perhaps the L lens badge is reserved for lenses that can be used for the entire range of EOS cameras unlike the EF-S series. Additionally, all new and future L lenses will sport weather sealing gaskets but digital SLRs with a 1.6x crop factor don’t have such seals so it doesn’t make any sense to put weather sealing in the EF-S lenses. The new EF-S lens may not be built to L standards of ruggedness, but based on the pics, I think they won’t be too shabby. Anyway, back to the optical evaluation.


Consider these other factors. On the traditional EF lens mount, wide angles have to employ a retro-focus design, due to the huge gap between the rear element to the camera’s film plane. With EF-S lenses however, designers are given the luxury of having the lens’ rear element positioned much closer to the film plane, making it easier to correct for optical aberrations. The EF-S lens also has a much smaller image circle and a medium speed max aperture, which again make it easier for Canon’s designers to correct optical flaws such as aberrations and light fall off, etc. It’s a fact that high speed, retro-focus wide angles with large image circle coverage is every optical designer’s biggest nightmare!

As far as flare and ghosting is concerned, I wouldn’t be too worried since it’s based on the same optical formula and the lens coatings are optimised for digital sensors. If anything, since the elements are fewer and smaller, I believe this lens may even be more flare resistant than the EF16-35L.

Canon also knows they’re up against some stiff competition. For example, Nikon is supplying relatively affordable DX zooms boasting SWM, ED, Asph elements and Vibration Reduction. Surely Canon will not want to be left behind by producing lenses that are inferior to the competition – especially for the desired target market of advanced amateurs who are the most dedicated “pixel-peepers.”

Finally, look at the price of the EF-S 10-22. It’s in the same price bracket as the EF17-40L lens which gives me confidence that quality glass is being used.

So, what’s my prediction? I think users of the EF-S10-22 will be pleasantly surprised. If this lens proves to be an optical gem, then Canon has poised itself to strike hard at the prosumer market with their slew of 1.6x crop DSLRs. Canon users who cannot afford DSLRs with larger FOV will actually benefit if this lens is good; finally, a true wideangle solution and increased reach for their tele lenses.

I’m keeping my fingers crossed that this prediction is spot on. What do you think?


Cheers,
Jonathan Kwok

[mpenza]

It is much lighter though at 385g vs 500g for the 17-40F4L. Hopefully, there won't be any vignetting and minimum distortion.

[oeyvind]

Not sure if there's any mistake... looks like EF-S 10-22mm USM has fluorite and 3 UD elements from I can I see here

http://www.canon-sales.co.jp/camera/..._22_f35_45.swf

but then from the text it says it has 2 GMo aspherical + 1 replica aspherical & 1 Super UD elements...

Quote: EF-Sレンズ対応カメラに新しい撮影領域をもたらす超広角ズームレンズが登場しました。EF-Sレンズの特長であるショートバックフォーカスと小径イメージサークルの利点を十分に活かした光学設計によ り、135フォーマット換算で16-35mm相当の焦点距離イメージとなる超広角ズーム域を実現しました。光学系にはスーパーUDレンズを採用 し色収差を良好に補正。また、GMo非球面レンズ2枚とレプリカ非球面レンズを1枚組み合わせて使用するこ とで、諸収差を良好に補正し、ズーム全域で高画質を実現しました。 2004年８月20日現在、EF-Sレンズ対応カメラはEOS 20D、EOS Kiss デジタルのみです。その他のEOSには装着できません。

[dkw]

Wow Jonathan, that was a lengthy post!

Thanks for bringing our attention to the use of exotic glass in the 10-22. If theory holds true, then this does promise to be of outstanding optical quality, falling short of 'L' grade on the basis of build quality (and market positioning)alone. Although the pricing is quite definitely 'L' league

Anxiously awaiting reviews.......

[kayaro888]

Good say Jonathan
Look at the spec, its sure better than the 3rd party lens like Sigma 12-24.
Price should not be differance too much, but Canon got wider angle and better speed (10-22 f3.5-4.5 vs 12-24 f4.5-5.6)

So which body to take: 1D2 (non EF-S) or 20D (EF-S)........

[perror]

Some people have mentioned that based on the MTF charts of the 10-22 that it should perform pretty well..maybe as good as the 17-40 in certain areas.. Hopefully that will be the case..

[oeyvind]

Just FYI: EF-S 10-22 will only be available in late Nov 2004 and EF-S 17-85 IS & USM version of EF 18-55 will be late September 2004.

[oeyvind]

Just FYI: EF-S 10-22 will only be available in late Nov 2004 and EF-S 17-85 IS & USM version of EF 18-55 will be late September 2004 together with the EOS 20D.

[Jonathan_Kwok]

Hey guys, the MTF charts on canon japan clearly shows that at least theoretically, the EFS10-22 drastically outperforms the venerable EF16-35L both in resolution and edge to edge sharpness.

This is truly good news indeed! Now, if only Canon can meet those theoretical MTF curves by ensuring tight manufacturing tolerances...I've seen way too many de-centered lenses, yes, even in L lenses, due to improper lens alignment in the production process... so much so that nowadays, I'll never pay for a lens until I've checked its optical performance and am happy with it.

A slight correction to my initial evaluation. I've just rechecked the lens diagrams and clearly the EFS10-22 is based more closely on the EF16-35L's optical layout than the 17-40. The EFS lens has one less element than the 16-35L due to the use of the Super UD which replaces the need for two UD elements. I think Canon has taken the best bits out of both the 17-35L and 16-35L and combined it into a small, high quality optical package in the EFS lens. Good move indeed.

What are its implications? Well, for one, entry level Eos DSLRS finally have a "kick-ass" superwide. 20D users will gloat in the fact that wideangle shots with this lens will outperform the results of wideangle L lenses mounted on the Eos 1D Mk II. I'm basing this statement on the assumption that the image quality between the two 8 megapixel camers will be closely comparable. Wouldn't it be ironic that the "lowly" EFS lens that may shame an L lens on the Mk II?

For people like me who love to hike into the wilderness carrying minimal equipment for landscape photography (ala Galen Rowell), then the 20D with EFS 10-22 will be godsent. The 20D and EFS lens combined will still weigh less than the 1DMk II body alone. And best of all, the results should be equal or superior to what the Mk II can produce with equivalent wideangle zooms.
The 1.6x crop will also mean an extended reach for tele lenses - great for those who may not have the financial clout to purchase a costly supertele.

I'm glad Canon did not cut corners with this lens (I hope I won't have to eat my words later). I'm glad they didn't base their design on the cheaper EF20-35 f/3.5-4.5 just to save some $$. This lens combined with the increased resolution and performance of the 20D makes it worthwhile for existing 10D users to upgrade.

For those of us who can't afford to buy the expensive 1 series DSLRS and need a high quality wideangle now, this lens can't be beat. And a few years from now if the prices of full frame DSLRs come down, one can still sell off this lens without too much loss. After all, it is relatively affordable and from what I can tell, 1.6x crop cameras will be around for a long time to come. Besides, even if you migrate to a full-frame camera, won't you want to keep a 1.6x crop DSLR like the 20D that will provide backup and extra reach for your tele lenses? Sounds like a great solution to me.

Cheers guys!
Jonathan

[Jonathan_Kwok]

OOps, here are the MTF chart links. The words are in Japanese, but you can understand the graphs anyway.

http://cweb.canon.jp/camera/ef/catal...22_f35_45.html

http://cweb.canon.jp/camera/ef/catal...6_35_f28l.html

Cheers guys,

Jonathan

[oeyvind]

Dude, it's a different scale all together! Plus one is 2.8, the other is not... anyway, it will works well in the right hands.

[Jonathan_Kwok]

Yes, I agree it's on a different scale. But that doesn't change my conclusions. The EFS is a superior lens.

The EFS10-22 MTF chart is based on a camera with a 1.6x Field of View (FOV) camera so the horizonal axis which mesures the distance from the optical axis to the edge of the frame is different (shorter). But this is of no concern to users because the EFS lens will ONLY be used on 1.6x FOV cameras.

Note however, that the vertical axis indicating lpm peformance is still using the same scale, meaning closer to 1 is better. The 16-35L is tested to cover a full-frame FOV camera - which accounts for a longer horizontal axis scale.

But look at the charts. Even if you were to mount the 16-35L on a 1.6 FOV camera, the MTF charts indicate that for the same area of coverage, the EFS lens is distinctly superior in terms of resolution height achieved. Even the MTF charts of the highly regarded 17-40L don't come close to the theoretical capabilities of this EFS. Which say a lot about its theoretical performance. Heck, the EFS MTF charts indicates that it even outperforms many wideangle primes.

I say "theoretical" because, actual production lenses will usually vary from these projected performance curves. But at least on paper, the EFS wideangle looks highly promising and is currently the finest performing wideangle zoom in Canon's arsenal.

Granted, I agree with you, one loses the f/2.8 aperture and weather sealing and it can only be used with 1.6x FOV cameras. But what you gain is closer minimum focus and it's about half the price of the 16-35L. At the wide end, it's merely half a stop slower then the f/2.8L zoom. In addition, since it has a tighter crop compared to full framed cameras, you'll gain increased depth of field for the same picture composition compared to FF cameras. Highly useful if you're shooting landscapes. Which means you can get away with a larger aperture for the same DOF compared to a full frame camera. Less motion blur in foliage blown by wind and don't need to stop down so much - less quality loss due to diffraction.

Please don't get me wrong, I'm not dismissing the EF16-35L's optical capabilities - I own one myself, but my point is that this EFS wideangle offers a truly affordable, superior wideangle solution which up till now did not exist in Canon's arsenal for 1.6x FOV DSLRs. And if these figures are true, and if the 20D's CMOS chip comes close in performance to the 1D Mk II, then the advantage will ironically go to users of the 20D with this lens. Wider angle coverage, higher resolution, better edge-to-edge sharpness, possibly lower flare, drastically reduced weight and best of all - a fraction of the cost. BTW, the EF16-35L has poor edge sharpness and that is reflected in the MTF charts. On a full-frame DSLR, that often means having unacceptable softness towards the edges, which will not be corrected completely no matter how much you stop down the lens.

At this point though, the EFS' optical performance remains purely speculation as the proof of the pudding is in the eating. I for one will eagerly wait to see how it actually tests out.

Ultimately, I concur with you...the final results still depends on the man behind the camera. But as far as comparing lenses are concerned, the EFS wide looks like one heck of a lens!



Cheers,
Jonathan

[oeyvind]

We can see you really like to write...

one thing I have to disagree with you is on the sharpness of 16-35, I do own one and I can tell you it's sharp edge to edge on what I have tested so far be in film, FF DSLR, 1.3x DSLR and 1.6x DSLR.

[Jed]

Originally Posted by oeyvind We can see you really like to write...

Agree! He is even more, er, comprehensive than me!

[Russ]

Keep it up Jonathan.
I am very keen on this lens!!

[RossChang]

Indeed very interesting lens tat serves well on a 1.6x... prehaps it would be more appealing if it were of L Lens Build...

If its really base on the 17-40, then we would probably be expecting the same, if not, similar optical attributes... which would be really nice...

Price?... would say it IS over priced baring the fact tat it's not L Built, and it's a EF-S

[Prismatic]

And people really, you should judge the lens on ACTUAL performance. MTF charts are just a gauge, should be taken with a bit of salt, pepper, soy sauce whatever.

Lenses performances are not limited to just MTF charts. For example, ergonomics, colour fidelity(very important) and ease of use. There's no point getting a super-sharp lenses when everytime you use it, you get pissed over slow focusing or simply hard to hold in the hand.

And really to those who shoot in less than ideal lighting conditions all the time, a F2.8 lense is a life-saver.

[ST1100]

i must say those 10-22 MTF graphs are really impressive for a wide angle zoom, even if goes only to 13mm from centre and is a slower lens.

[RossChang]

I never knew how to read MTF charts...

does it really matter compare to hands on testing the lens?

[aaronng]

Just to let you know, MTF charts only show readings at infinite focus. So the 10-22 will be sharper than the 16-35L at infinite, but at regular working distances, the MTF is not valid.

Aaron