Are We Analyzing Small Details Too Much?

(commentary)

Ever since the D4/D800 combo was announced (and now with a D600 in the middle in terms of pixel count), I keep getting the "which is better" and "do I need" questions. I haven't posted my reviews of the D4 and D600 yet (hope to by the end of the month), so these questions repeat, but now with details, such as: "the DxOMarks are 96, 94, and 89 (D800E, D600, D4), what does that mean?"

It means something and it means nothing. I've long been a proponent of testing your equipment and understanding what it can and can't do. Nothing's changed there. But to be frank, we long ago passed the point where small differences in test results are meaningful to most folk.

Let me illustrate. Let's say that I took my D600, D800E, and D4 to a basketball game and shot them side by side, then had Sports Illustrated use a photo I took in their magazine. Could you tell which camera I used for that photo? I'm betting that the answer is no for 99.9% of you. Even the 0.1% might not be comfortably sure, and they might actually be keying on something that isn't particularly important in that shot to make their determination.

Which camera would I have preferred in that instance? The D4, of course. It's better suited for the type of shooting I would have been doing (action). Technically, the D800E would have produced slightly better acuity, all else equal, but with the dot gain in the printing process, I suspect no one would actually see it. Likewise, if Sports Illustrated used the image on their Web site, it probably would be limited to 800 pixels or less. Even assuming we had to crop to the final image, that's not a high enough bar that the difference between DxOMark's 96, 94, and 89 numbers would show up in any way.

Don't get me wrong, there are times when a particular camera is "right" and another one isn't, based on the sensor. If your goal is print landscapes big, get a D800E. The D4 would frustrate you. This has little to do with dynamic range or any of those other measurements that people are getting overly anal about, but simply with pixel count: you have more sampling with 36mp than you do with 16mp.

Many readers of this site were aghast when I dropped my DX kit for an m4/3 in the backcountry. How could the OM-D E-M5 beat a D7000 for landscape work? Well, since I generally don't print that work larger than 16mp allows, there's not enough difference for me to worry about. True, I tend to use filters a bit more with the E-M5 than I did with the D7000, mainly to pull in dynamic range, but that's an example of my understanding what the tests I did mean in a practical sense: yes, the D7000 was better by a small bit, but I could do something about that. Technically, the Olympus lenses I'm using are better than some of the Nikkors I was using, so there were trade offs both directions.

Thus, I test to understand the limitations of my equipment, then adjust my shooting accordingly, not to pick equipment. But it's surprising how little I have to adjust sometimes (the first SI example I use, above).

Many of us are using Marianne Oelund's Optical Simulator these days to try to understand how specific parameters impact very low level (pixel and sub-pixel) results. It's definitely an interesting scientific pursuit, and I've learned a few things from it along the way. But are any of those extremely low level things impacting my choice of camera? No. They only help me understand my camera of choice at a very low level.

So here's the scoop on FX these days, based upon not agonizing too much over the low level details:

    D3/D3s -- I'll take the D3s over the D3 because there was a difference in sensor that has an impact on high ISO work. But both are still very good choices for a lot of the types of work where you need ISO 3200 and ISO 6400. You have to ask yourself if you really need more than 12mp. If not, nothing wrong with using these cameras.

    D3x -- It's even more clear with the D3x, though with a minor caveat. The D3x is a remarkably good 24mp camera with an exceedingly high body build. Unless you're pushing into very high ISO ranges (the caveat), you don't need a D600 or D800. Of course, Nikon overpriced the D3x ;~).

    D4 -- It's very much like a D3s, only with 16mp. Thus, if you need more megapixels plus the other things a D3 does, it's a valid choice. If you don't, stick with your D3s.

    D700 -- Same as a D3, only in the smaller body style. Just like a D3, there's nothing particularly wrong here if you don't need more than 12mp. You can agonize over getting a new camera, or you can save your money and get a new lens instead. Which is going to net you a bigger gain: >12mp or better lens?

    D600 -- As the DxOMark numbers indicate, the low-level differences aren't huge between the three current Nikon FX bodies. Thus, the D600 is the bargain camera of the bunch. But it has fewer features and a lower body build than the others. Can you live with that to get a bargain?

    D800 -- You buy this camera if you truly need more pixels. Maybe you buy this camera if you are using smaller images but need more acuity. But you give up fps and you'll need more RAM and drive storage, so you truly have to need the pixels to go this far.

    D800E -- The winner on acuity and resolution at the moment. Landscape shooters printing big probably ought to land here, but you'll be needing good skill sets and a really good lens to fully take advantage. And remember, you need more RAM and storage if you're going to go here. Also remember that if you're dealing with static landscapes, you might be able to get to this point with your current camera using stitching.

If you didn't quite get what I said: I'm choosing between these cameras primarily based mostly on feature set for intended use, secondarily on pixel count needs. Some of you will probably throw cost into the decision hopper, too (e.g. a used or refurbished D3s might be a good choice over a D4 if you don't need 16mp). But these are all high-level specifications that are easily distinguished without testing.

Here's what I don't agonize over these days, but it seems many of you are still splitting hairs on: dynamic range. Every camera I just mentioned has enough dynamic range for me. Every camera I just mentioned captures more dynamic range than you can print directly. Virtually all of my landscape work these days compresses dynamic range (not linearly or evenly; much like I was doing in the film days). Fill and graduated NDs are important tools, as are post processing contrast tools and techniques that deal with pieces of the dynamic range at a time (some of you may remember that I supplied an action with my earlier books that split the captured range into 10 "zones" (layers in Photoshop) of contrast, for individual manipulation).

Don't get hung up on numbers and tests. Try to understand whether the camera helps you do what you need it to do. This little article is bracketed by "Good Design, Good Images" parts one and two. Analyze whether the camera you choose has a design that helps/lets you make the images you choose to make, nothing more. If all you're ever doing is making Facebook snapshots to share with friends, a D800E shouldn't be your choice ;~).

Virtually all of my responses to people asking "which" all boil back to that: understand what you're trying to do, find the products that let you do that, then choose either the least expensive one or the one you like the best. That's it. And yes, there will be multiple products that let you do what you want/need. Comparing low level numbers on them to make the choice is not the right thing to do. Which is cheaper? Which do you like best? (After handling it in a store, preferably.)

There's a really good example of that happening in the mirrorless world right now, for example. The Olympus OM-D E-M5 and Panasonic GH3 are within whiskers of each other on the numbers, but the E-M5 tends to be the winner by very small, actually teeny, margins. Margins that aren't meaningful to selecting which one to buy. Price, size, menus, features, ergonomics are all much more meaningful things to make the choice on. This is one of the reasons why I don't publish the numbers I get from my testing any more. Over reliance on those is delusional. If a camera is one Rat's Eyelash better than the other on Test XYZ but you always press the wrong button because of the design and miss the picture, who cares about the Eyelash? If the camera that's an Eyelash better is one you can't afford, who cares?

Be pragmatic in your decisions about gear. Make sure that the things and design points that will impact your shooting most are what you're basing your decisions on. Truly, how many of you can say that 11.7 stops of dynamic range versus 11.2 is the thing that will impact your shooting most? Comprende?

Every FX camera Nikon has ever made is a winner. At something. For someone. Every FX camera Nikon has made. Every one. You'll only lose if you do the analysis of which you need wrong, and those low level numbers are probably the wrong thing to analyze, so stop obsessing on them.

(Aside: OMG, Thom wrote something positive with the word Nikon in the sentence. Is Thom sick? Yes, sick like a fox. ;~) Get a grip people: I write positive comments about positive things. I write negative comments about negative things. Nikon's QA and customer support in 2012 were (mostly) not positive and were a significant story of the latter half of last year. Let's hope for a better 2013 in that respect. Stop shooting the messenger. Note that one of the stories below is that NikonUSA set up an online parts store, as I suggested they should. It solves some, but not all of the problems that I reported on. Messengers are sometimes good things ;~)

Looking for gear-specific information? Check out our other Web sites:
mirrorless: sansmirror.com | general: bythom.com| Z System: zsystemuser.com | film SLR: filmbodies.com


dslrbodies: all text and original images © 2024 Thom Hogan
portions Copyright 1999-2023 Thom Hogan
All Rights Reserved — the contents of this site, including but not limited to its text, illustrations, and concepts, 
may not be utilized, directly or indirectly, to inform, train, or improve any artificial intelligence program or system. 

Advertisement: