Archived posting to the Leica Users Group, 2002/10/03
[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]Hi Tom, > You noted that you have tested lenses of various types and you > did not see a > difference. Without any background about the test, the test metric, and > test equipment, doesn't that represent the same type of blanket, self > serving statement that you so often accuse others of using? To put it simply, no. My statement was not a blanket statement at all. It was purely limited to my experience, which I clearly stated was only my experience. I said that I've tested them and have not found any difference, that does not mean any difference does not exist, or other testing did not result in different conclusions. > You wrote: > /*****************************************/ > You can adjust the color fidelity of film same > as you can digital, so I don't see how that claim holds. > /************************************************/ > The first part of your sentence, prior to the comma, is > sophistry. The part > of the sentence following that comma is true. Obviously, we have different definitions of "fidelity". It, without definition, can be a rather amorphous term. I believe it is generally accepted that fidelity, as we are using it in this discussion, is "the degree to which an electronic device accurately reproduces...". There are many aspects to color "fidelity" (much less image fidelity) and the major ones in my book include: 1) accurate reproduction of "A" particular color 2) accuracy of encompassing the overall spectrum (gamut, overall range) 3) "resolution" of color differentiation within that "spectrum" 4) accuracy of relativity between colors. 1 and 4 are adjustable, as I said, and you "disagreed" with. 2 and 3 would be considered "lost data" (or unattainable data)...but, is losing this data in this regard important... You can certainly decrease 2 and 3 via software adjustments, which, of course, would degrade fidelity. If any of the properties I've described above are adjustable, and it increases (or decreases) the accuracy of the reproduction, then the "fidelity" has changed. Given this, then my statement (prior to the comma) is entirely true and not "sophistry", as you so claimed. As far as the overall spectrum, perhaps there is an edge for digital...and that's debatable from comparing the two response curves, and what the human eye can perceive, as well as it's usefulness given the limitations of output medium. My understanding is the human eye spectral response is ~360 to ~760nm, and both curves "more or less" encompass that range...but for the film curve, we don't see how the response tails off, unfortunately. This particular issue may or may not be relevant, as the output device seem to have a more limited output "gamut" that are not capable of encompassing the two ends of the scale. The spectrum of colors that can be viewed by the human eye is wider than any method of reproducing color (monitor/printer etc.) that I am aware of, though you may know differently...so you have to do something with the colors that you captured that are outside of the range of your output medium...if you wanted to output them, and that, IMO, could very well reduce fidelity. You would have to somehow compress the spectrum into the output medium limits, or remove the data outside the printable spectrum. If you simply compressed the spectrum, you'd get inaccuracies over at least some part of the range, simply because you can't move colors around without getting overlaps (combining them), unless you move adjacent ones such that they don't overlap. Obviously, there are different choices of how to handle this, but, as I said, it would reduce output fidelity IMO. The most "accurate" thing to do is simply match the output colors to the input colors, and either clip the ones that don't match, or you can simply combine them with others...loosing "distinction" in those you combine with, and those outside the range. As far as resolution within that spectrum...well, both can resolve to a much higher level than the human eye can discern (or can be output), so that, IMO, is not an issue...but, film has the edge on that. Because a digital camera has 12 bits, doesn't mean all 12 of those bits are actually good or even used. That is very dependant on the design of the camera. If there is an adjustable analog gain stage (actually two, one for each setpoint) that allows the adjustment of the setpoints such that the input signal from the CCD can be "expanded" to match the A/D input voltage range, then you can get 12 bits, providing you're not into the noise...but if there is no such analog stage, then you are limited as to what you actually get, as it will occupy only a portion of the 12 bit range. And...then there's that nasty old Bayer pattern used in one-shot digital cameras (except Foveon equipped cameras) that I'll discuss below... <snip> > Digital scanning products with visual color matching function > response, have > been delivered to the packaging and graphic arts markets for well over 20 > years. These products offer excellent color reproduction and require far > less processing to match original colors in print applications. Most > importantly, the provide useful color information out side of the > reproducable range of film See above. I do not believe that is conducive to "fidelity", which is what we were discussing, as it can distort the accuracy of other colors, depending on what "method" of incorporating them into the output medium you use. > and they allow specific areas of color space to > match up better with printable gamuts so there is a lot less > correction that > needs to be done. The range of color for the output device is less than that of the input device, so what good does mapping colors that are outside the reproducible range of the output device if you are trying to retain "fidelity", which, by definition means accuracy. This may be where the problem lies...in our differing definitions of fidelity. > If you look at the Kodak site and download some of their sensor literature > > http://www.kodak.com/US/plugins/acrobat/en/digital/ccd/kaf16802ce.pdf > > you will see that the RGB sensitivity functions overlap quite a > bit. It also overlaps with film as well, though, unfortunately, they don't show the tails in the spec sheets I've got, but they are there in reality. > These > sensors can capture the full spectral range of the human visual > system Well...hum...so can film... > with > out gaps. Film captures it's range without gaps as well... I understand what you are saying, but without the tails of the film curves, it's really impossible to tell, and second, how much significance it really has is debatable. > The reproduced colors may be off, but they are still captured. CCDs don't do well with red. Also, there is the issue of the Bayer pattern used in one shot imaging sensors (except Foveon). The actual color resolution is severely reduced, as these sensors only have %50 green, %25 red, and %25 blue sensor elements. The data is then interpolated, which means made up, yes, algorithmically, but it still is not actual accurate image data. That is certainly not conducive to "fidelity", now is it? Obviously, scanning backs do a much better job, and don't have this issue...but we were talking about one-shot cameras, as scanning backs are really only useful in the studio...which isn't where most people use their Leica I would venture to guess. Also, all this, and we haven't even broached the digital vs film B&W issue...which I'd prefer to save for a later date. <snip...> > The goal was never to prove a point, only to > learn what was right. I'm all for that. Sometimes on the path to understanding what is "right" though, points have to be made, and in order for them to be accepted, and to some degree of reason, proven. > The imaging world is filled with a lot of marketing hype. It's a sad > reality. As imagists, we need to share our experiences and each > of us must > try to relate those experiences to our own context. Agreed, and...it should be done in a "respectful" manner. <snip...> > I don't give a big rat's ass about educating the world, I'll > leave that to you, but for goodness sakes, hold yourself to the same > standards that you hold other contributers to this list. I do. Your post, whether you meant it to or not, came across to me, at least, as being rather "smug". I certainly don't mind discussing things, being questioned, and even being mistaken, but please do so without the pontification and smugness. I know quite a lot about this subject, and I know I don't know everything about it, nor did I ever say I did. And you may very well know a lot about this subject as well, and it's pretty certain you also don't know everything either. Perhaps you ought consider taking a bit of your own professor's advice... Regards, Austin - -- To unsubscribe, see http://mejac.palo-alto.ca.us/leica-users/unsub.html