...Is not identical to Ctein's. He speaks for himself.
My own position might be too subtle for its own good, because I've talked about it before but it doesn't seem to penetrate very far out into the world.
I did a lot of work years ago comparing film formats experimentally; we made a lot of prints and questioned a lot of people about them, both photographers and non-photographers. Generally—remember I'm saying generally—the results conformed to a broad pattern. Imagine a bar chart:
- - --------------- - -
- - --------------- - -
- - --------------- - -
- - --------------- - -
- - --------------- - -
(I'm sorry about the above—I'm not a graphic designer, and it would take me forever to render that in a more elegant form.) The bars represent formats, the x-axis is image quality (IQ), and those are "fuzzy" ends to the bars. What I found, more or less, was that film formats (the "sensor size" of the film era) had a certain amount of overlap when mapped against enlargement size. Each successively larger film size had either no advantage or only a subtle advantage over the next smaller format up until the print size reaches a certain rough threshold. (In many cases this was experimentally observer-dependent; photographers could discern differences that non-photographers simply couldn't see. I suspect that's the same with digital.)
Digital is similar, except there are at least three more complicating factors obscuring a clear result: ISO, image size (pixel count), and sensor generation. By that last term, I mean that it's useless to generalize between 2002 and 2010 cameras, for instance.
Another complication is that many people—even some photographers—look at only some factors of IQ and prefer to overlook others. For instance, most people love to look at resolution, and it's historically been very popular to look at shadow noise. There are of course many other pertinent qualities of a digital file, from color accuracy, to the transition to highlight clipping, to the robustness of the raw file (how well it accepts correction or modification in post processing). The variables add up quickly, making strict experimental distinctions difficult even before you start considering that many different photographers have different aims or intentions for their work and how it will be used. Whether or not people make prints—and how large they'll go if they do—is one huge variable in that respect. Finally, different people have different tolerance levels for the various faults; I've never been much bothered by noise, for example, to the point that in the past I've been amused at how fanatical some people get about it. This is the "YMMV" clause, and it's important.
Ctein's general rule of thumb that you need a 50% change in image size—number of pixels—before you'll see obvious qualitative differences is useful. I have an equivalent belief about sensor size, although I wouldn't put a percentage value to it. What he's saying is that you need a fairly large difference before the difference becomes readily apparent, and that's what I'm claiming here too. There simply isn't that much difference between APS-C and full-frame, all else being equal. And there's even less difference between APS-C and 4/3. And yet there is some...see those overlapping bars, above. When you compare 4/3 and full-frame, however—analogous to bars two places away from each other—the bars are further offset from each other, meaning you start to see more easily detectable differences more often than not.
Peepology
Last night, Imaging-Resource (I-R) posted some of its early test shots from the Nikon J1 1-System camera with its CX sensor. Looking at the results in I-R's fascinating Comparometer (which essentially allows you to pixel-peep any two cameras' results side by side) leads to a fascinating but still wholly predictable conclusion:
CX vs. DX: J1 at ISO 1600 is not too far behind the D90 at the
same ISO. An impressive result for the new camera, really.
DX vs. FX: D90 also holds its own pretty well against the D700.
CX vs. FX: Compare the J1 to the D700, however, and the qualitative
differences become more obvious.
Note that you should really go to the Comparometer yourself to really see the differences; the smaller versions posted here aren't really adequate, even after clicking on the illustrations to make them bigger. (Also, consider making a small donation to I-R if you pore over their Comparometer files for any length of time—all that bandwidth really does get expensive for them.)
Note also that, although I've tried to control for image size in this illustration (the J1's 10.1 MP vs. the D90's 12.3 and the D700's 12.1), I've had to violate another variable: the D700 and D90 are 2008 cameras, and the J1 just came out in 2011. The comparison that is fair with regard to sensor generation will be complicated by different image sizes, as the 2010 D7000 is 16.2 MP and no one expects the "D800" (D700 replacement) to be a 12-MP camera when it eventually gets here.
Still, it shows the general trend: to consistently see a fairly distinct difference in image quality, it's almost necessary to move two sensor sizes from where you are, not one.
He just isn't going to, is all
Finally, although I don't want to speak for him, I believe Ctein's contention—or should I say half of his contention? That size doesn't / does matter—is based on base ISO. My qualified agreement with him is also predicated on base ISO. Even my 2009 Panasonic GF1 begins to falter at ISO 800, and digicam sensors fall off in quality a lot quicker than it does. I would never recommend a digicam-size (what I call "fingernail size") sensor to a pro. In fact, Ctein seems to have forgotten this, but several years ago I tried to convince him to replace his Fuji with a proper DSLR! I gave up because he convinced me he wasn't going to do it, not because I became convinced that it didn't matter.
For the record, I still think he should get a Pentax 645D, the closest digital descendant of the Pentax 6x7's he used for many decades. Or at least a Canon 5D Mark II.
To wrap this up...
There are so many variables with regard to inherent digital camera image quality that it's almost necessary to try to get a "holistic" or overarching "feel" for the many variables and differences and how they balance. As always, it's quite possible to work backwards, too: that is, you settle on a camera and then, after exploring its capabilities, you use it within the areas where its competencies overlap with your tastes, expectations, and requirements. We photographers have been doing that since the beginning, and I don't mean the beginning of digital.
Mike
Send this post to a friend
Please help support TOP by patronizing our sponsors B&H Photo and Amazon
Note: Links in this post may be to our affiliates; sales through affiliate links may benefit this site. More...
Original contents copyright 2011 by Michael C. Johnston and/or the bylined author. All Rights Reserved.
Ctein responds: There's really not a lot of difference in our positions about this.
I'd have two qualifiers—these days I think of "base ISO" as being anything up to about ISO 400. I just haven't noticed much degradation in image quality among any of the better cameras with increasing ISO over that range. It's all down on the pixel-peeping level.
ISO 1600 still seems to be the great divide where things really start to shake out.
Funny thing is that's a lot like the way it was with film. Makes me start to wonder if we're seeing technology differences or marketing-think differences, as in "Well, we don't think your typical small format (film/sensor) buyer really cares about ultra-high ISOs." Just wondering....
Another quibble—I think the generation thing needs more weight than you give it. We're not close to theoretical maximum sensor efficiency yet (give it another ten years), so there can be really big differences between makes and generations. When most potential buyers think about this, they don't compare apples and apples, time-wise. They're mentally setting the latest small-sensor announcement against the larger format camera they already own or have been thinking about buying.
So, I think more often than not, when folks think "this small sensor camera can't possibly compete with a large-sensor one" what they're really thinking and don't realize it is "this brand new state of the art small sensor camera couldn't possibly compete with a 1–4 year old large sensor one."
Technically, they should be comparing apples and apples. Consumerly (is that a word? should it be?) I don't think they do, so that becomes a more realistic framing of the discussion.
You're not wrong about me and the Pentax 645D or the CaNikon 5DmIV/D3s.
Problem is, you're entirely right. Either way.
About a year ago I thought I might be making a quick sale of my whole Jim Marshall collection, so I was thinking about what camera to buy. Had the collector who was sniffing about bit, I'd have been able to buy any camera I wanted, even a Leica. So I started thinking about what I wanted, if money were not an issue. It boiled down to a list of 4–5 different cameras, all of which were equally desirable and which were entirely different from each other! What that told me is that I didn't know what I really wanted. (I know I don't want to own 4–5 different camera systems, that's for sure!)
So, issues of folding cash aside, I decided it wasn't time for me to commit. Still doesn't feel like it.
Featured Comment by Bill Pierce: "One of the biggest arguments for larger sensors is larger prints. Museums and galleries tending to exhibit larger and larger prints of contemporary work may be a pretty good argument for some photographers to use larger sensors."
Mike replies: Right you are. See here, for instance.
Of course, many of us do not exhibit in museums and galleries, but that doesn't negate your point for everybody.
Featured Comment by John Camp: "To me, all those results look 'plastic.' [Attempted humor alert.]
"I write books for a living, and my publisher wanted to update the author photos, so they hired a West Coast pro named Rob Greer for a photo shoot that took place Wednesday. We do this every few years, so I'm pretty familiar with the procedure—but the shoots are spaced far enough apart that that I started with film, then did some joint film/early digital stuff, and the last three times, it's been all digital, but of different generations.
"Rob was using a D3x with, I think, a Nikon 70–210 zoom for the outdoors, natural light shots. Before I went over for the shoot, I shaved, and noticed I had a small ingrown hair on my chin, making a red spot. I plucked it out with a tweezers (too much information?) and then went off to the shoot.
"I have a theory, which I just compounded about thirty seconds ago, that you're never more critical of photographs than when the photos are of you. When Rob sent me the photos, the first thing I saw was a little, tiny, red dot on my chin. I couldn't see it in the mirror, but it jumps right out of the large D2x files—and I'm not talking about head shots here—you can see it clearly in the three-quarters body-length shots. It'll go away in the final Photoshopped pictures, but it really brought home to me exactly where we're at, and how nonsensical some of our discussions have become, as regards resolution and sharpness.
"The spot on my chin was smaller than the head of a straight pin (I mean, it was the base of a whisker.) Exactly what would be the purpose of doubling the resolution on this? I can think of a few places where you might want this—on a space telescope, perhaps, or if you were taking technical photos of fabrics, but I have to say that further resolution (which we're apparently about to get) seems to me to be driven by engineer-think: we can do it, so let's do it. But, as you (Mike) say, there are a lot of things to be balanced in photography and it seems to me that this obsessive drive for resolution has to be stealing from other things, like high ISO capability, dynamic range, or pleasant shoulders, or whatever. Or, maybe, that's all the developers have got: resolution. But look what happened to Kodak's films when Fuji began kicking out specialty films with more saturation, different levels of warmth, etc. Suddenly, everybody was shooting Fuji. Is it possible that one of these companies wild decide that there's enough resolution, and we suddenly get a generation of sensors with some other qualities?"
Featured Comment by Hugh Crawford: "Regarding the differences between 2008 and 2011 camera: the advances in software are just as dramatic. The same raw file from my 1Ds in 2005 looks radically better rendered with 2011 software compared to what it looked like in 2005. Even a three year old image re rendered in this year's Lightroom has a world of difference. I suspect that if the comparo-meter re-rendered the files every year, the results would quite different."
Featured Comment by Pak-Ming Wan: "That's a pretty good analogy on how the sensor makes such a difference. Whilst in Australia a couple of years ago, I borrowed a D700, and I was astonished at the files it put out. Astonished. If you have ever used a top-tier sensor, you ought to; just to know the difference. Then afterwards, and only afterwards, are you able to make a real decision on what kind of sensor you'll need. Me? Micro 4/3, GF1."
Featured Comment by Michael Bernstein: "I'm really not a big fan of I-R's Comparometer. True, it is convenient. But unless you make a habit of shooting JPEGs with near default settings, it is a very misleading way to compare the abilities of different sensors/cameras. Certainly, some manufacturers are more conscientious about optimizing JPEG output than others. But that's not how anybody who actually cares about pixel-level quality works."
Mike replies: Yep, you know what they say—nothing's perfect. We haven't even started to take into account the crowd that likes to compare different raw converters and how they match up to different cameras—when I compared the D3, 5D Mark II, and A900, I worked from raw files, but I got slammed for using the same raw converter for all three cameras—the one I used supposedly showed one of the cameras at a disadvantage, according to some. So, what, then—they wanted me to first try each camera with multiple raw converters, and learn the ins and outs of each raw converter I chose to make the best possible conversions, then pit the cameras' IQ against each other's? Someday when I have a free month, maybe. I gotta come up with new material here every day!
There's really no such thing as a completely neutral test when there are so many potential variables. The Comparometer still beats having to go out and buy all those different cameras yourself to make the comparisons from scratch.
(
), 173 copies
(not available in the U.K.), 32 copies
(
), 145 copies
(
), 198 copies. British publisher Thames & Hudson isn't shipping to the U.K. yet, so that's U.S. only. Aperture is the publisher here. (Incredibly, I couldn't get Aperture to email me a few public relations JPEGs for our article—even though they told me they have them prepared and ready to go. I even made a toll call to their PR Department to request them. Web publications get no respect, I tells ya, no respect at all.)
, (
), 31 copies.
