I’ve been writing about focus accuracy for some time now. No lens
test is any good unless it’s checked and cross-checked, and it’s one reason that more and more I use field shots in judging lens performance. Field shots
or “3D” shots are also hard to do when comparing cameras/lenses, but they are the reality of making actual images, which is what the end game
is. Choosing one lens over another demands the ability to consistently achieve good results; it doesn’t matter if the lens is world-class if (for
example) it can’t be focused consistently.
The guys over at slrgear have lens testing down to a
science, inventing their “yellow submarine” to do focus bracketing. That’s something I’ve been doing for years now, but not having
the lab space I do it manually, cross-checking results with different cameras, field shots, etc. There are surprises and lenses perform differently at
different distances, so field images really matter.
Another problem is that the sharpest results in the center aren’t necessarily optimal for the frame as whole; this can be seen
shooting a series of focus-bracketed shots, where “giving up” a tiny amount in the center can yield a superior overall result. Few of my images
rely on center sharpness; most are off-center.
Field curvature assumptions are also a problem. The Zeiss ZF 28/2 Distagon has enough field curvature to make a very noticeable
difference at the 5.5 feet cited in the slrgear article. See my Dec
29, 2007 blog entry (rice pounder) for an example where the corners are sharp, in spite of being several feet behind the center.
The bottom line: field shots of a wide variety of subjects are still the best way to evaluate lens performance for real images,
especially if your shooting involves certain distances or types of subject matter. That is what I do in my Zeiss
ZF Lenses review. The myriad effects a lens produces (the “look”) are often what matter the most in terms of visual impact.
Canon EOS 5D Mark II commentsPERMALINK
In yesterday’s entry, I called Canon’s 1Ds Mark III and 5D Mark II “lukewarm” but didn’t explain what I meant. One reader questioned
why I called it the “Mark Time”, a term which I now realize suggested my disdain for the 5D Mark II, which is not at all the case—I want one!
I meant the “Mark Time” comment more generally as applying to any brand, but I did not make that or the reasons clear.
As I write this, I see the Nikon D3 and Canon 1Ds Mark III sitting on my desk to my right. Were I to grab a camera for all around
shooting at any given moment, it’s the D3 that I’d pick up 80% of the time. But why is that?
When I make a satisfying 12MP image with the Nikon D3, its 12MP are good enough for my needs most of the time. It’s also
about low-light performance or even the cleanliness of the image at ISO 200. I
like shooting at dusk and for that the D3 is superb. There are so many good photographs
to be made that have little or nothing to do with 12 vs 21 megapixels. And when megapixels matter, the lens quality must be there and the focus must be
perfect.
So by “marking time” I mean that we get more of the same (mostly) with the EOS 5D Mark II, Leica S2, etc, not a
leap forward in sensor technology such as 30-stop dynamic range (true or firmware based), elimination of de-bayering (true color), no anti-aliasing filter,
monochrome options, etc. In short, we’re not getting the choices we could be getting (yet), because the mainstream focus is still marking time
on megapixels—that’s what sells these days. And I’m not too unhappy about that, as I like megapixels as much as the next person.
There is one big exception to foregoing with the Canon 5D Mark II, and that is the convergence of video and still images. Offering
1080p video in the 5D Mark II is a big reason I’ll want to get one. I’ve always disliked having to carry a separate tape-based camcorder, and so I rarely
do. The utility of combined video and stills has always been there, just poorly executed (crappy stills on a good camcorder).
The 21 megapixels of the EOS 5DM2 is a huge crowd-pleaser (myself included), but the practical advantages of 21 vs 12
megapixels exist mainly in theory for many uses/users. Most DSLR users (who aren’t pros or even serious amateurs) would be better served by a wide dynamic
range 6MP full-frame camera offering the ability to shoot at 1/500 sec @ f/5.6 in near darkness. But even many pros would like such a camera.
Which brings me to my first theoretical dream camera: The NiCan DR6 and DR6m, offering a 6MP full-frame sensor with 16-bit dynamic
range and images at ISO 1600 with the noise of the Nikon D3 at ISO 200. Along with a monochrome version
offering ISO 6400 at the same noise level.
The second dream camera, acknowledging the megapixel advantage for some applications, is the NiCan DR42, offering a full-frame
42MP sensor with ultra-clean ISO 25 and 50 for those tripod-based high-res shots. Low ISO and low frame rate would be just fine for shots needing that
kind of resolution.
I used to shoot (and still have) a 4X5 view camera. If you’ve never seen a well-executed 4X5 “chrome”...well it’s a thing of beauty.
An 8X10 chrome is even more gorgeous. With large film formats, we have the best of two things: impressive tonality as well as impressive detail.
Which brings me back to the Canon EOS 5D Mark II: it’s a nice step in the right direction, but at 45° to the target, which is
high resolution with all the other goodies: wide dynamic range, low noise without banding or artifacts, focus stacking, elimination of lens aberrations,
etc. My “marking time” comment simply means that optimizing some of those other areas while staying in the 16MP range would be a tradeoff I’d welcome.