[Zig]

Forgive the moniker: search-engineering, dontcha know...
Herewith something a little more in-depth concerning the above 2 lenses. As it includes several images I've spanned it across a few threads.
I lack the resources, wisdom and insight to be thorough and exhaustive: I've compiled a comparison of sorts but hardly stratospheric; I hope it may prove a useful resource or merely be of interest. My findings are based on my full-frame experience. Time of writing is July 2010.
For a Canon-user, searching for the Holy Grail of wide or superwide angle lenses can be costly and frustrating. Making the assumption that one wins the lottery of(allegedly ) Canon's quality control and has deep pockets, options include 14mm and 24mm primes(both fast and excelling even more in their MkII incarnations).
We then have the 17mm f4 TSE and the stunning MkII version of the 24mm TSE: stonkingly good, tailored for architectural/landscape work. I will not explore the fors and againsts of these: autofocus, use of filters, flare control, etc; there are rafts of opinions and findings out there by more skilled and better-researched chaps than me.
Canon's wide-zoom, the MkII 16-35mm F2.8L(and which I owned until last week when I sold it for a Zeiss prime), has attracted both plaudits and howls of frustration, depending on who one listens to..and, I suspect, how long one has owned it.Very sharp centrally, a photojourno's everyday tool, all-round landscape lens and also useful for portraits...yet with many a description of "mushy" edges and corners, which means cropping away a good many pixelsworth and composing a shot with this drawback in mind. On a full-frame camera, this can often(until one has gone through a few copies) be accompanied by uneven sharpness across the frame even at optimum apertures. And yet, in the one lens you also have excellent central performance from 16 to even 28mm:and unless you get a poor copy of it up against a good copy of Canon's 17-40L, I believe it to be the superior lens. Of course, if one is needing, say, 17mm at f4, then the 16-35 wins hands down.
Canon users thus have ever been keen, if landscapes are their bag, to stray outside the fold.
Little temptation has been needed over the past couple of years. Previously, the devastatingly good Zeiss Distagon 21mm f2.8 had rightly attracted the sobriquet "legendary" in hushed tones. Reportedly, in the Kyocera/Contax days, only 5000 of these were made. As befits a legend, rumours of just why its quality was/is so stellar, have been rife: that shimmer, that irridescence...was it down to "banned" ingredients such as arsenic and cadmium in the glass?
Since Zeiss has brought out a series of its lenses in Canon(ZE) fit, though, it seems that despite the loss of aperture ring and the inclusion of another element, the quality is still rare and peerless..and of course at a price: yet such concerns are ephemeral for such brilliance. Also in ZE fit is the 18mm Distagon: reportedly very good indeed, though with the occasional report of uneven sharpness across the frame and a tendency to exhibit chromatic aberration in the corners. For those lary of these lenses now being made in Japan, bear in mind that in many of the places that dismissively mention "Cosina glass", the same noses are upturned for some reason as they say, "ahh..shame about the moustache distortion...".
And then there's the Nikon 14-24mm G.
No aperture-ring and no communication with a Canon body at all. At time of writing, in fact, there is globally just the one adapter that successfully allows, via a lever mechanism, to choose aperture: reportedly, this works well...but a note of caution: a programmable EXIF/focus confirm chip along with tax will, for a UK user, push the combined cost of the lens and adapter to almost the cost of a Canon 24mm TSE MkII. I don't want to, er, wet on anyone's strawberries here, but another note of caution: if you are a IDs user(that is, pre-MkIII), you'll need another gubbins as well from the site that sells the adapter. Only trouble is, there is no option in the add-to-cart to order this. I have read of many disgruntled people who have tried in vain to email the developer, only to receive no reply. One 1DsMkII owner has so far waited over 18 months for his adapter, and his Canon 1Ds and Nikon lens have yet to couple, as it were. Also, for several months, a link to the version 1.3 pdf pertaining to the adapter still only delivers v1.2. Attempts to contact the creator have met with varied degrees of success.
I myself actually ordered this fine lens, confident that this adapter was a done deal; as I've a 1Ds MkII, however, I developed cold feet.
Consequently(and so the reason this comparison has come about), I sold my Canon 16-35 and bought a Zeiss Distagon 21mm instead.

(For those who want to cut to the chase and see further shots with this spectacular lens: Shaun Reeder: Zeiss T* Distagon 21mm f2.8 ZE)

This was not without trepidation. Reduced to a cold sweat by reading mtf charts(and who hasn't), I saw that on paper the relative figures were so close...was I getting rid of, effectively, at least 3 "primes" just for one that would match the performance of the optimum spot of the 16-35?

And, I've always found the 16-35 Mk2 to be no slouch; heck, at 21 to 24mm and at f11, the centre-to-edge definition was(I thought) as good as it gets.
Or, should I say, "got".

It's a year or two between these 2 shots, though I'll summarise the similarities:
The "old"(Canon) and "new"(Zeiss) shots were taken at the same time of day, though with the old one later in the year. Though not an exact match, they are pretty close: the Canon is at 24mm, the Zeiss(obviously) at 21mm; the Canon is at a very sharp f9, the Zeiss at f11. Both shots were taken on a IDsMk2, both converted at low-contrast settings in Adobe Raw to 16-bit, uncropped tifs. There was no sharpening of any kind added at any stage. The resulting 100% crops were finally converted to 8-bits jpeg, all at level 9 in CS2.
The Canon lens had a high-quality polariser attached; the Zeiss did not. Both uncropped tifs started out at the "front end" with a size of 4992 x 3328 pixels, this being a 16.7 MP sensor.
Finally, note that the Zeiss shot could be said to be "handicapped" by being shot at ISO 400; the Canon was at ISO 100. OOO-er. :/

Firstly, here are the similarly-converted "straight" shots, straight from raw, downsampled as above:

The Canon 16-35mm f2.8L MkII; 24mm, f9, 1/40s



And the Zeiss Distagon T* 21mm f2.8 ZE; 21mm, f11, 1/100s



AlllllllRIGHTYthen: remember not to draw any conclusions at this stage from perceived colour: different seasons and conditions provide differing colour temperatures..and both the polarised Canon shot and "flat" low-contrast conversion will queer the pitch further.

In the next parts(which I'll do tomorrow, as I'm wiped out), I'll post the all-too-revealing crops of corners, edges and other detail, (which will announce their own conclusions), before summing up and offering subjective comment.
More tomorrow!

[Toad]

Another fascinating missive. I am going out on a limb here and predicting a win for the Zeiss - primarily based on eyeballing the bottom corners in these small un-cropped (and un-zoomed) images.