If the PMA 2005 show taught us anything, it’s that photography retailers have a lot of explaining to do this year. Promoting and demonstrating digital cameras has always relied on a few familiar, comfortable old assumptions like “DSLRs are more expensive than fixed-lens cameras,” and “more pixels produce higher resolution,” and “resolution is what determines how good a camera is in the first place.” With those concepts firmly lodged in our minds, and more importantly, in our customers’ minds, we could coast along through any presentation, introducing sales points as mindlessly and reflexively as we enter our PINs at ATMs. But at the recent PMA show, we watched the grim truth jell. DSLRs are more expensive than fixed-lens cameras, for example, but now only sometimes. As discussed in detail here last month, there are now at least four DSLR models intended to sell for “under $1000.” So much for that easy answer to “what camera should I buy?” And then there’s the pixels issue. Has that ever been as simple as we’d have liked it to be?
Fujifilm was among the first to make it complex. Sometime back in the twentieth century they announced something they called the Super CCD, offering it in their first DSLR since the early nineties. The FinePix Pro S1 had a 3MP imager at a time when 3MP was the going figure among most digicams. But Fujifilm said there are 3MP cameras, and there are 3MP cameras. Their 3MP camera, with its Super CCD, was a better 3MP camera because its pixels were shaped like polygons, and arranged in diagonal rows. And everyone nodded and smiled, and waited for them to continue. But that was it. Polygons and diagonals. They said their piece, and sat back down.
Well, there was more to it than that. The S1 and its otherwise similar successor, the FinePix Pro S2, indicated that although their Super CCDs possessed 3 and 6 million pixels respectively (in rounded numbers), they could output pictures closer to 6MP and 12MP respectively. Smart software tricks were the explanation, but not just your old-fashioned pixel-doubling or interpolation. For starters, although the enlarged picture had nearly double the pixels, it wasn’t exactly double. Fujifilm representatives said they thought a better term was “extrapolation.” Everyone nodded and smiled again, but the Fujifilm representatives sat back down.
Unfortunately for most of us, the mathematics of digital code are about as easy to express in language as architecture is in music. You can make metaphors and allusions, and spin rhetoric all day, but when you are all done you will have left something out.
Well, if you can’t explain it, you can replace it with something else. How about “everyone knows that”? As in, “Polygons and diagonals are good, everyone knows that.” As long as the Super CCD produced a superior picture—or even a merely good one—everyone would eventually know there’s something good about polygons and diagonals.
But is there? That’s the rub. Camera reviewers rose to the challenge of finding out, and how do you find out? By designing tough tests for the product. Some Fujifilm representatives say the tests were tougher than would be made on normal cameras, the bar raised because Fuji’s claims suggested it ought to be. Which sometimes left the S1 with a black eye.
But judged on the whole, cameras with Super CCDs kept receiving high praise for design and picture quality, from modest consumer models to the pro DSLRs. The repetitive dull thud of these findings began rising in chorus, becoming a mantra for wizards and gurus: “Polygons and diagonals are good.”
Who cares how it works, as long as it does?
The RAW Beef
Three or four years ago, mostly from sources at Kodak, we began hearing a lot about the “digital negative” created by the image sensor, and how it was “destroyed” by in-camera processing and JPEG compression. The best picture obtainable, at the full 36-bit color and with a broad range of contrast, is what should be preserved in the archive. Squeeze it down to 24-bit color, boost colors here, brightness there, till you get the output—the print—you like best. But save that original, because some day you may want to do it differently (new printer, say).
In a market where the pixel-count was collapsing as an argument for sales, and manufacturers were looking for features by which to tout their cameras, there was little argument against the unprocessed—or RAW—camera file. Sure, it completely bypassed what TruePic Tubo and DIGIC were all about, which was a specific treatment and look of the output, but who cares? It’s an option. Digital photography’s all about options.
One of the strongest arguments for the unprocessed picture came from Foveon, or rather Sigma whose SD9 camera introduced Foveon’s X3 imager. The sharpness of that camera’s output quickly became legend. Its use of a RAW format was not the only reason, but it was part of a package that made for exceptional quality.
Meantime, all cameras are festooned with great features, all of which can be turned off. Why shouldn’t the processing be turn-offable, too? Practically everyone these days—not just DSLRs, but prosumercams as well—offer a proprietary RAW format, and proprietary software to read it. The trend has become so prolific that Adobe has introduced a new RAW format, .dng, which they recommend all other RAW formats be converted to. It could improve the chances that ten years from now, there’ll be a way to open all those proprietary-format pictures.
What’s the truth about RAW? Is it as indispensable as its advocates say, or an unnecessary diversion from computer models that come close to perfection? There are probably fundamentalists on both sides of the question. But we’re moderates, and we like everything.
A RAW file has all the good qualities its supporters claim, but sometimes those qualities exceed the demands of the scene being photographed. What if it’s an overcast day and the lighting’s flat and the colors are mostly grays? Do you need 36-bit recording for that? Do you even need 24-bit? The RAW file gives full user control over dynamic range, all 9 EVs of it in many cameras, but in the diffused light of a cloudy day, does the contrast exceed 4 stops? Is there anything you would do in these conditions that is better than what the in-camera processing system would do? Is it so much better that it’s worth the effort of adjusting the image? It takes only a couple minutes per picture, but hey, you shot a lot of pictures today.
There are those who might say it’s heresy, but as good as the RAW file is, it’s sometimes better than needed. There are times when a processed file, even a JPEG, will produce as good a picture as anything can from the materials it starts off with. This is not to say the JPEG would perfectly match the photographer’s taste. It might need further adjustment to bring out, say, a certain section of the frame the photographer thinks should be prominent. Nobody said you can’t further adjust the JPEGs processed by the camera.