Many years ago I heard about a friend of the family who was a semi-pro sports photographer. As he was a trained optical technician, he would collect a handful of new lenses from the local camera shop, test them all, and then return them all apart from the best sample which he bought. I think he shot Nikon. I don't see many shops allowing this.

Interesting. I did the very same thing when I purchased a new 100 2.8 macro at Best Try last year. However, before I purchased (and to avoid a "restocking fee"), we unboxed the two units they had on the shelf and shot some samples with a tripod. Checked the shots via lightroom on a laptop and I walk out with the best of the two. It was quite educational for the sales reps and a few customers, we had some fun. I use that lens for all my table-top products.

I was just in there yesterday with the wife shopping for a new TV, and spotted a new D7 on the shelf, so I'll have to make it point to go back and do some 'testing' on that one.

In the last few months I've made it a hobby of mine to purchase old vintage lenses from ebay and experiment, (taking them apart, cleaning them up and shooting). Surprisingly, I've found a few gems for pocket change, along with some absolute dogs of course, but certainly more interesting than falling asleep while cruising forums, and that has revived my shooting slump.

My point is, that when I find a lens worth bidding on. I do some research online, and then compare my results with others opinions. Interestingly, my results were mixed. Meaning that there are simply too many factors involved to believe that any particular lens will perform as good, (or as poorly) as anothers own experience. Variances in expectations, usage, (including age for used), and lens quality controls can add up, and you're left with the one and only conclusion. Buy it, and then try it.

There are no universal methods and standards for lens testing. as far as i know DXO and Imatest (used by Photozone, dpreview - and myself) are the practical choices for testing of which Imatest is the cheaper and more accessible. Imatest can be used in a wide variety of ways on a wide variety of test targets and can generate a mountain of data which you need to wade thru to reach a reasonably informed conclusion. Needless to say, the ultimate data is influenced by good practice in choice of target, processing of the digital image in-camera or in computer (e.g. sharpening directly affects resolution numbers), choice of target, choice of data sampled, etc. etc. Lots of choices judgements involved just as there are in "subjectively" evaluating comparative images. And hence the caveat from imatest that the numbers generated are not absolute, but of comparing different lenses evaluated under the same conditions. Anyone who's looked at Photozone data vs. dpreview data (or any other) will immediately see that Photozone is reporting numbers which are not comparable to dpreview.

That's the preamble to my testing over the last few days of 2 100 L IS lenses. Since the bottom line is whether or not to replace the 100M, i shot it as well and the 90TS and 200 f2.8 (recentlly returned from Canon service) under the same conditions using the Imatest SFRplus target and Imatest software. And i easily generated a pile of data from which i'm trying to report a sensible abstract.

Micro adjustment for both of the 100 L lenses was -3 on my 5D2 (the 100M and 200 f2.8 were -2). Tests were shot using autofocus (except the TS of course) with liveview as a convenience for framing and mirror lockup with the 2-second timer and IR remote release. I did most testing in direct sun, but confirmed with a second indoor test in diffuse light (with remarkably consistent results). Imatest recommends illumnating the target with lights at 45 degrees - this probably makes some difference to the absolute numbers but it's a nuisance not not related to the way i'll use the lenses.

The bottom line is that there is not much to choose from between the two samples - one is centered slightly to the right, the other to the left (the 90TS with it's big image circle appears to be perfectly centered and the 200m nearly so - the 100M is slighty de-centered) but resolution, distortion, CA are otherwise vitually identical - these are the highest resolution lenses i've tested. At f8, 15% sharper in the center than the 90TS and 100M and 6% sharper than the 200 f2.8. I tested from wide open to f22 (full stop increments) and my results were consistent with dpreview (although my absolute numbers were different). I compared lenses at f8 for the best balance of center and edge performance, but there are no surprises at other apertures. As you stop down to f11 and beyond, diffraction starts to equalize the lenses with the 90 TS holding up slightly better at f16 and f22.

The main "issue" with the 100 L's is CA which unfortunately appears to interact with the Imatest target and software - at least as i'm using it. If my math is correct, CA for both R/C and B/Y is about the same as dpreview shows for R/C. This is about 50% more than the 100M, twice the 200, and 8 times the 90 TS (which doesn's show visible CA at 2:1). The issue with Imatest is that resoluting is measured at as many as 23 vertical and 23 horizontal edges with the SFRplus target. When the Imatest software looks at the vertical edges the measured resolution appears to be limited by the R/C or B/Y fringing on the vertical edges. Looking at the vertical edges 35-45% to left and right of center, resolution is off around 8% -- at 78-85% out resolution is 30% down from center! On the other hand if you look at resolution of the horizontal edges, reslolution is down around 5% 35-45% out and 7% 75-85% out. For comparison the 90 TS shows essentially no resolution difference between horizontal and vertical edges anywhere, the 100M and 200 a few percent at the outer zone. My 100-400 has a bit worse CA than the 100 L -- and a greater difference between Imatest reported resolution of horizontal and vertical edges. Is this an inevitable "feature" of complex optics? And more importantly, how much does it matter unless you're photographing white picket fences against a dark background?

So looking at the resolution of horizontal edges, the 100 L IS is astounding across he image. Measured on the vertical edges, resolution at the left and right is no better or slightly worse than the other lenses. I welcome informed comment on this issue.


Autofocus - One of my main interests in the 100 L IS is improved autofocus. It does not appear to hesitate, hunt, or clunk like the 100M at close focusing distances, but it will take some experience to verify how much quicker it is overall.

Quirks and quibbles - The front element of the 100 L appears to be exactly the same diameter as the "old" lens even though filter mount has gone up to 67mm (styling? manufacturing economy?) so it looks like a reducer ring needs to be more or less permanently in place to take my 62mm 500D, polarizer, and macro lite adapter.

With the exception of CA, the 100L is a better and generally more useful lens and i'll replace the "old" lens. For macro shooting stationary subjects alone it would be hard to justify, but then i'd rather use the 90TS with liveview on a tripod for those subjects anyhow.

"Autofocus - One of my main interests in the 100 L IS is improved autofocus. It does not appear to hesitate, hunt, or clunk like the 100M at close focusing distances, but it will take some experience to verify how much quicker it is overall."

That's the point. It depends on what you intend to use the lens for. I'm not replacing my 100mm as I use it for studio, and prints from it are sharp enough, and that any nuance or slight different will be lost in the process. Frankly, at f8 with studio flash, there are a multitude of lenses, vintage included, that will render sharp enough for prints that reach the limits of my sensor. 80% of what I publish is digital and online anyway, and even so, designers have little issue with using 100% crops of my shots as brochure insets or call outs without having me to re-shoot.

In fact, I've tested some old FD lenses via an optical adaptor and compared them directly to their EF counterparts, and aside from a stop of light and the extended magnification, they are completely useful for my needs. If you don't believe me, go here: http://ferling.net/galleries/StudioEFvsFD/index.html

Granted, in the field, with available light, the IS alone would be reason enough to upgrade, and I would not be too concerned if the results were no sharper than my non-IS, non-L version, because that's not the point. The point of IS is to regain some resolution that you would have lost while moving about. We look at numbers and charts to understand the 'possible' resolution a lens will have. But we don't have those numbers or a chart that represents the ultimate degradation of that 'possible' resolution our when usage deviates from the test bench. To say that a given lens is no sharper than another is a true statement, but only for that very particular lens. If you could have 100 samples at your disposal to choose from, then you might have a more accurate answer.

Yes, and I'm very happy with my new 100 IS L. It's excellent even at wide aperatures.

However, just out of curiousity, I put my Nikon doublet (corrected, 2-lens) closeup lens on my 90 TS to see how they'd compare at about a 10" distance. At f/11 they are the same aside from the macro having a flatter field. In casual flower shots, they are the same. Given the importance of correct focus and the tilt, I'll actually be using my 90 TS with a closeup lens for some uses.

Paul
www.PaulRoark.com

Over the last week I've had a project that required me to shoot with both my previous generation Canon 100 Macro and the Canon 180 L Macro. The difference is significant on a full frame camera. The 180 makes anything from the 100 look soft by comparison.

It's seems reasonable to conclude that though a completely different optical formula, and focal length, Canon could easily improve the 100 by simply introducing improved glass. Add in IS and the potential for improved image increases.

What's unfortunate is that Canon's QC, or lack there of, has apparently allowed huge variation in what gets shipped to the consumer. You would think in introducing a new L lens they'd be all over QC. But then you'd also think easy mirror lockup on a new body would be a priority by now.

As I understand it, diffraction is a fundamental physical limitation that cannot be overcome by improved optical design, so it's only to be expected that the performance of different lenses tends to equalize as you stop down.

I have taken few shots in a two Monoblock set up with my 1D III and EF 100mm f/2.8L IS just to test shooting HSS with Profoto compact 600's with TT1/TT5 combination, and I have found the lens to be impressive.

I also, as I have said in an other thread, notice more and more copy variations in Canon's lenses lately, not very good unfortunately, but testing is required, as we can't expect all lenses to be the same, unfortunately! :-(


Here is one of the shots, please note that nothing was done to the pic, and it was a test for the HSS.