This was by no means a scientific or particularly comprehensive test. We wanted to see how the Sigma SD14 and the 14MP Kodak stacked up purely in resolution/detail. We did not test for colour accuracy, dynamic range, corner darkening or colour fringing.
Basing the tests on an analysis of actual prints and displaying the findings as scans of the prints rather than pixel-peeping is unusual in the testing world but it has certain advantages in that we did not need to worry about re-sampling and size matching. It is also more representative of how cameras of this class are likely to be used. I think it is the way to go, I hope you do so, too.
So what do we think overall?
My opinion is that in print sizes up to 24 x 16 inches (possibly larger too, but we did not test this) the SD14 and Kodak 14nx make prints that are essentially indistinguishable. I don’t think there is a lot more to say really! If I was being absolutely nit-picking I would maybe suggest that there is a tiny trace more detail in the Kodak files but I can’t really say what size print would be required for this to be visible. Several metres across, I would guess.
I don’t personally own a SD14 or Kodak 14nx but I do own a Canon 5D and an earlier generation Kodak 14n. I don’t consider there to be any practical resolution difference between these two cameras, so by extrapolation I would be confident in saying there is little or no resolution difference between the SD14 and the Canon 5D that you would notice in prints of this size.
Are Sigma therefore justified in advertising the SD14 as a 14MP camera? In practical photography, viewing prints at typical enlargement sizes, its resolution holds up against the 14MP Kodak, so in that respect I suppose they are. But personally, I’m still uneasy about their unilateral re-definition of the pixel. However, in view of the currently unsatisfactory method of pixel counting and the tenuous relationship between MP counts and real resolution and faced with the alternative, calling the SD14 a 4.7MP 3-layer sensor – which really doesn’t really give the right impression of the quality of image it can produce – reluctantly, I’ll probably have to go with the Sigma definition. If you are thinking of buying a SD14, you would be advised to do any necessary comparisons yourself and obviously there are issues other than final print quality to consider; but I’m happy to recommend it on print quality.
I agree for the most part with David’s assessment. That is, the 13”x19” inch prints are virtually identical. The same goes for the 16×24 inch prints. I think in one case the pine needles at middle distance (20 to 40 yards away) look a little better from the Kodak. But once you move pine trees closer or farther away this advantage disappears. Also, broad green leaves seem identical as well.
For what it’s worth, the Kodak files look a bit better at 100%. But this really has no bearing on visible print resolution, which is what I was interested in. But if you wanted to say the Kodak or Canon 5D had higher resolution and looked a bit better at 100%, I would say that’s a fair statement. What would not be correct is to conclude from that statement that a slightly better 100% crop translates into a better print at 16×24 and smaller. It doesn’t.
I should point out that I was looking for a difference between the two cameras that one could classify as: startling, clearly visible, incredible, slam dunk or any other phrase that would indicate a clear, unambiguous victory of one camera over the other. Much like medium format could easily defeat 35mm. It was, more or less, an astonishing tie.
All photos were taken on a tripod. Auto focus and manual focus was used to eliminate both system and human error. (But then again, is it really that hard to tell if a photo is in focus or not?) Sigma was shot at f/8 using the 50mm macro lens. The Kodak was shot at f/11 and the lens used was the Nikkor 85mm f/1.4. Upsampling for both was performed using the S-Spline XL algorithm in PhotoZoom Pro 2.0. The dpi was 300 for both inkjet and lightjet prints. The SD14 file required a bit less sharpening than the Kodak.
I think what all this work really shows is that there is a sad disconnect between the real world performance of actual prints vs. pixel peeping 100% crops. Obviously people feel comfortable looking at 100% crops. They think it’s a quick metric they can use to gauge imaging systems. Of course one can always object and say that all prints do is show you the strengths and weakness of a printing system. This misses the point that ink on paper (nowadays) is indeed what a photograph is, especially from DSLRs. It is not an image on a screen that disappears when the electrons stop flowing. It is something tangible that you can view with photons from our own sun at any time you like. It’s something you can hang on a wall, pass around to friends or keep in a box under your bed. It is, I say, the ‘thing’ you should look at when considering any camera.
This is not really part of the main test but Eric has asked for some additional images to be included. This link goes to an appendix page that includes some scans sized for a more accurate representation on screen of the size of the crops on the print; a scan covering a larger areas than the tiny crops and an additional crop from one of the prints that demonstrates the Kodak does have an advantage in rendering some detail such as pine needles.
NB The content for the appendices was lost following a change of hosting service. Sorry!