|[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]|
Phil-- I'm finding it hard to follow your train of thought, and you seem to be using "profile" differently than I understand it from an ICC standpoint, so let me seek clarification on a few points: At 3:48 PM -0800 18-3-02, Phil.Lippincott wrote: >Bill >You have mixed metaphors between reflective scans of film profiles >and film profiles. BF: I don't understand what you mean by the above. We've been discussing two IT8 specs, one that applies to photographic prints and another that applies to photographic (positive) transparencies. Both specs start out by saying (in essence) that two dye families can produce a color patch that the human eye will perceive as identical but an electronic scanner will perceive as different. Whether the light that hits our eyes (or a scanner's sensors) were reflected off the target or transmitted through it is irrelevant. So I don't see where the mixed metaphor comes in. >Yes all IT8 targets are suppose to represent only the human eyes >perception of color with a particular iluminate e.g. D50 or 5000 >degree K. BF: True, the IT8 targets are by design keyed to human perception under a D50 illuminant. >Also yes some scanners not many can capture the entire density and >energy spectral response of film e.g. drum scanner with a density >response between .02 to 3.9. Each photomultiplier can see the whole >spectral response, but there is a filter in front of each PMT, so >the scanner will return an RGB data stream. BF: As far as I know, all electronic scanners in commercial production do as you describe above: use filters to sample the intensities of light in three bands of light and return R, G and B values. Were you thinking of my comment that "when scanners are spectrophotometers..."? By this comment I meant to contrast the current "primitive" state of the art where we measure intensities of light that hits the sensors after passing through three reddish, bluish and greenish filters then guessing at the spectral distribution of the spot we're sampling with some hypothetical future when we can actually measure the intensity of each visible wavelength and thus "know" the spectral distribution of each spot. >It is also true that CCD scanners can't see as much density as most >film targets because the best CCD max out at ISO Status A density of >3.0; and film targets go to 3.3. Thus the CCD scanners clip black. BF: Of course. >If you are profiling a device dependent input profile for a scanner... BF: By "profiling a...profile" do you mean "creating" one? > ...it's purpose doesn't have to be to correct for the media you are >using, it can be to simply convert it's data to a profile connection >space. BF: To my way of thinking the purpose of a scanner profile is to know, for each RGB value it outputs, what "color" was shown to it. (And to oversimplify, you create a scanner profile by showing it a bunch of known colors and recording the RGB values it reports for each one.) The value of having a scanner profile is that when you get (from a scanner) a collection of RGB values representing the pixels of an image, you can use the profile to figure out what "real" colors those values represent. Now that you know the "real" colors of the source image you can use this knowledge in a variety of ways: color correction, converting to other color systems or color spaces, archiving an image, etc. >An also interesting point to consider is that as IT8 targets are >respresenting human perceivable color that also means that they do >not give a full spectral representation of the density of the film >they are printed on. BF: That doesn't seem quite right. Patches 0 and 23 (unnumbered patches to the left and right of the greyscale at the bottom) are supposed to be the minimum and maximum densities the target media can support, and it's specifically noted that these extremes might not be color neutral for a given dye family. >For example if you read a Kodak Q60 with a densitometer you will see >the second to last spot on the target for the blackest gray is >Status A density 2.7 and the Black is 3.3 This is over an entire >stop of exposure left off the target. BF: Step 22 (next to blackest) has a specified value that is (as you've measured) not as black as the media can support. I think this was chosen as a very dark patch that all media are supposed to be able to attain. The blacker, unnumbered patch to it's right (patch "23") is specified to be the Dmax of the media. Thus how far these two patches are apart in density will depend on how much denser the media can be than the value specified for patch 22, which is supposed to be the same density no matter what media (or dye family) a given IT8 target uses. >Yet the assumption is if you are trying to calibrate a scanner you >don't need accurate shadow calibration greater than 2.7. It's OK to >interpolate in software whatever happens between 2.7 and your >scanner black because the human eye won't care. This maybe somewhat >true, but many of my clients do care and find that IT8 targets are >not good enough to get density accurate scan to print. BF: I, too, want to be able to more precisely characterize my scanners in this range (I'd also like to be able to afford scanners that can actually "see" well in this range ;-). >An ICC workflow can [work?] (if you are willing to accept the >density range of your targets) for closed loop reflective to >reflective profiled workflow. BF: You seem to be saying that a scanner profile can be no better than the target used to make it, which makes perfect sense. But what do you mean by "closed loop"? The traditional meaning of "closed loop" is that you choose one set of products (photo paper, scanner, printer, inks, etc.) and adjust everything so that what you print at the end consistently looks like what you scanned at the beginning, but only if you use the same equipment, the same process, the same materials, etc. My understanding is that ICC profiling is designed to open the loop; so that "any" scanner can produce a file that can be used on "any" imaging workstation to prepare it for printing on "any" printer and still get consistent, predictable results. (I put "any" in quotes to emphasize that there are still stringent qualifications that each item must meet for this all to work.) >For instance this can allow you to profile the scanner then blind >scan that profile for other materials within the gamut of your >profile to then print to your profiled output device. BF: Huh? >The color temperature, gamut limits and other caveats previously >mentioned all apply, but it works. > >Best Regards >Phil Lippincott Bill Fernandez wrote: > As the IT8 spec explains, a given scanner will interpret the colors > of each dye family differently, and that the purpose of each variant > of an IT8 target is to represent the gamut and characteristics of a > given dye family so that a scanner calibrated (and/or profiled) to > the target will give consistent results with other images from that > dye family. > > The IT8 target is NOT meant to make it so that you can toss anything > > (watercolor, magazine, oil painting, phographic print, slide, etc.) > > on a scanner bed and get the same colors out that your eyes saw going > > in. This is impossible to do with a single profile against a single >> target using current scanner technology. >> >> When scanners are spectrophotometers and are able to measure the >> intensities of hundereds of different wavelengths across the visible >> spectrum (rather than three bands in the reddish, greenish and bluish >> portions of the spectrum) things will be different... >> >> --Bill >> >> > >- >Turn off HTML mail features. Keep quoted material short. Use accurate >subject lines. http://www.leben.com/lists for list instructions. -- ====================================================================== Bill Fernandez * User Interface Architect * Bill Fernandez Design (505) 346-3080 * email@example.com * http://billfernandez.com ====================================================================== - Turn off HTML mail features. Keep quoted material short. Use accurate subject lines. http://www.leben.com/lists for list instructions.