So, I'm at the camera store yesterday chatting w/ one of the employees there, and he mentions that there was a published study (ostensibly on the Internet) comparing all of the various sensors out there for various factors, and it ends up rating them for sharpness (possibly among other things - that's about where our conversation got cut off). He also said that the results were heavily slanted towards Nikon - like, it had the top sensor (D2X) and the third sharpest, as well, and Canon didn't show up in the results until 5th or below.

I can find no such study via multiple searches via Google, etc - of course, I might be searching on the wrong terms. Also, I'd have expected the Foveon sensor to be in there at the top, and he didn't even mention that, so...

Is this guy feeding me a load of crap??? (which I suspect, but am willing to give him the benefit of the doubt... briefly... )

Ah yes, camera store salesman. A most reliable and reputable breed.

To my knowledge there has been no such study. In any event, any camera with an AA filter (such as all Nikon and Canon models) has inherently lower sharpness than one that doesn't, such as all medium format backs, and the Leica M8, for example.

The whole thing is a phantom in any event since so-called sharpness is a function of a wide range of factors, including the sensor, presence or lack of an AA filter, lens used, and so forth.

Michael

...and then there's the question of practical significance. The average Joe doesn't have any use of a Ferrari being able to speed up to 300km/h in 0.5 seconds (nitpickers don't need to bother) when the speed limit between the home and work is 50km/h.

The point being, even if you have the sharpest lenses and sensors and whatnot, if you only post 800x600 images on the net, or print 20x30cm papers, a bit of added sharpness is insignificant. It's all about perspective and context.

I was struggling not to roll my eyes at him in the store



This was an important point, to me - if the study were true, I really wanted to see how they claimed to mitigate all of those factors!! I figured I'd better double check my facts, though, before I wrote the guy off as a loon...

Henrik - I've made some very nice 20x30" (not cm) prints from my 10D... for whatever that's worth... if that's not getting the Ford Taurus up to 300km/h, I don't know what is Haven't tried it w/ my 30D, yet, but... The 10D starts to fall apart at that size, and might not pass muster for everyone, but they seemed pretty reasonable to me, given the amount of enlargement... But, point totally taken

I'll hopefully get a post up today that makes clear why I even bothered to have a conversation with the gent at all... he was a nice guy, but I could tell we were going in a bad direction... hell, all I wanted was a Nikon system brochure...

In that case your question is reasonable, no doubt. Although the following is old news and you most probably are very much aware of this, I never tire to remind people that large prints require a farther viewing distance, which should be taken into consideration. Sure, some prints are droplet-peeped upon, no matter how gargantuan they would be, but, as you already claimed, even the 10D manages those dimensions quite well, just as long as the viewing distance is greater than just a few.

Still, I'm convinced (partly by personal experience, partly by the general opinion), that those improvements in quality that seem significant on the monitor (especially on the TFT), are hugely smoothed and minimized once the photo is on the paper. Even Nikon gets a load of dirt for its noisy sensor, the noise on paper is often insignificant or even enhancing to the image (the illusion of added detail).

I'm not trying to teach you (there are many other better suited for this than me), I'm just reciting for everyone else interested in this thread

I actually experienced the opposite - one ~80DPI enlargement showed the chroma noise way more than what it looked like on the screen, particularly in the blue sky. Very disappointing. It was clearly visible even at a reasonable viewing distance. I do agree, though, that luminance noise can give a nice effect -- or perhaps that's just because we're used to similar-looking grain on film.

-Lars

I'm always open to learning, though... teach away!!!

Such a comparative test was done a few months ago by a private poster and images posted for comparison. I cannot find the link back, but the d2x was indeed performing surprisingly well both in terms of sharpness, lack of moire and image uniformity. Although it has 4 MP less, it was ahead of the 1ds2 for instance.

The difficulty though is with the raw converter. Most universal raw converters are optimized for one camera, while proprietary raw converters might have different levels of sharpening applied.

What it would take would be to gather 10 experts, one for each system, and ask them to unleash their skills to produce the best possible image from a given scene.

This was never done as far as I know. Besides, there would never be an agreement on the interest of such a test, since landscape shooters would have different priorities than portrait shooters for instance... The bottom line is that there is no best system, only one that suits your needs best.

Besides, you might want to look into the whole pixel peeping debate to understand better why such comparisons are done or not done.

Cheers,
Bernard

Forget about this kind of sensor info. I'll settle for a simple but comprehensive list of senor sizes, their semiconductor technologies, their manufacturers, and whose sensor is in each camera model.

In a way, sensors are equivalent to film types. In this context, when one buys a digital camera, one is buying not only the camera body, but is also committing to a particular film type forever. I miss the good old days when Kodak publishes volumes of technical info on each film they made.

I must confess, I've been something of an equipment hound in the past - I seem to recall Michael having coined a term for this sort of person a while back, but its name escapes me However, along the way, as I've grown more comfortable w/ the technical aspects, I've sort of reversed philosophy to the point where I could care less about what technology a sensor is made from (CCD, CMOS, ABC, whatever) - I want to see what kind of pictures it can make. After that, I'll consider the technical angles - like CMOS generally consuming a lot less energy, etc (though even that doesn't make the only difference in camera battery life...).

ETA - it just struck me as odd the way this guy talked about this "report", so I figured I might investigate, in case I'd been totally asleep and missed something like that... I realize, now, that my query about it here should have been a self-answering question... heh heh...

I like the analogy to film types, though - even though we can manipulate the crap out of an image, if we so desire, its still got a basic set of information to start with - a "flavor", if you will - and that's never going to change.

In addition to this D2X I'm going to be playing with shortly, I'm also getting quite a large amount of medium and large format gear (I'm a bit overwhelmed, actually) - I found myself poring over Kodak and Fuji film specs again I'd forgotten how much I enjoyed reading those... as if it made a difference to my photography

Maybe he is talking about sharpness in the traditional "absolute" sense used to measure and describe film sharpness, like MTF at a given number of lines per mm, or the highest l/mm at which anything at all can be resolved, regardless of format size.

Such measures of "sharpness" in l/mm are probably heavily influenced by pixel spacing (though effected by other factors too, like AA filters and in-camera JPEG conversion and sharpening choices unless one does one's own raw conversion). So I am willing to guess that the rankings were something like
- Nikon D2Xs (5.5 micron pixel spacing)
- Olympus E-500 (5.4 microns, but a bit "softer" due to stronger AA filter and/or more conservative sharpening of JPEG output)
- Olympus E-330 (5.6 microns), but possibly too recent to have been tested
- Nikon D200 (6 microns), and/or Sony A-100, Nikon D80 if not too recent to have been tested
- Canon 20D, 30D and/or 350D (6.4 microns)

Such test results probably need updating now due to new models like the Olympus E-400 at 4.7 microns and the Canon 400D at 5.7 microns, and a bunch of recent 6 micron options.

Why on earth would you consider facts like semi-conductor technologies or sensor manufacturer of more relevance than direct measurements of performance characteristics?

Strange: you are interested in such information about films, but you just said that you are not interested in the equivalent information about sensors. Anyway, the good news is that Kodak also publishes lots of technical details of its sensors:
http://www.kodak.com/US/en/dpq/site/SENSOR...iesRoot_product
and in particular
http://www.kodak.com/US/en/dpq/site/SENSOR...meProductFamily