Re: filmscanners: Fw: Color Profiles for Scanners
Hi Shaf, Tony. shAf wrote: Tony writes ... On Fri, 19 Jan 2001 22:28:59 + Photoscientia ([EMAIL PROTECTED]) wrote: CIE L*a*b* is poorly suited to DTP applications in three main areas: The main objection to CIELAB for DTP is (AFAIK) that DTP is (professionally) always done in CMYK for pre-press output - eg Pantone is CMYK gamut - and CIELAB is an RGB space. I would object to L*a*b, as any application space, simply because it is unintuitive. My own questions regarding it are only in the context of it being a absolute reference ... otherwise (practically) useless. Hear hear, Shaf. BTW those words about CIELAB and DTP weren't mine. They were extracted from a well argued paper on the subject by a very knowledgable colour scientist. I would add, though, that no matter how accurate a reference or not CIELAB is, what it certainly isn't is portable, or practically useful to the average user. How many people can lay their hands on a CIE reference light source? Then, having got your light source, now what? There's no way to reconstruct a colour from it's L*a*b* specification, apart from trial and error, mixing standard pigments with a CIE calibrated colorimeter, and of course, we all have one of those handy in our back pocket. You don't? Pity. But I suspect you have a perfectly useable colour reconstruction device sitting right in front of you now. An RGB monitor. Forget L*a*b*, you need an entire lab(oratory) to make proper use of it, unless, of course, you convert it to RGB first. Now in defense of RGB: The phosphors in monitors aren't all different. They're usually one of a standard set. The first thing I do with a new monitor is to set up the colour balance in hardware (this involves taking the back off, in most cases). I match the white point to north daylight by eye with the gun gains, and adjust the RGB black levels for a neutral greyscale, and just invisible raster. The settings interact, and it's time consuming. Even after a lot of practise it still takes up to half an hour, but what it doesn't take is a roomful of special equipment. I've found that a monitor set up like this by eye alone will match almost any other monitor set up the same way, within very close limits. These monitors aren't used for any special graphics purpose, just workstation clusters, but it's nicer to see a row of monitors that all look the same. So what's the point of that anecdote? Well, if the manufacturers could be bothered to do the same thing, using a standard electronic substitute for my Mk1, then I don't think there'd be half the moaning there is about RGB spaces. In fact, I'm sure it's not beyond the wit of man to make all monitors self-calibrating in some way, thus compensating for ageing, drift and user abuse. Then we could completely throw CIELAB out of the window, or Macos, as the case may be. As things stand, all we've got is half the equation, with the sRGB specification laying down standards that people just ignore. Regards, Pete.
RE: filmscanners: Fw: Color Profiles for Scanners
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Photoscientia Sent: Tuesday, January 23, 2001 4:41 PM To: [EMAIL PROTECTED] Subject: Re: filmscanners: Fw: Color Profiles for Scanners I would add, though, that no matter how accurate a reference or not CIELAB is, what it certainly isn't is portable, or practically useful to the average user. I don't understand why this is an issue. CIELAB is an exact mathematical transformation of CIEXYZ, which is the standard that ICC uses. This is all math, hidden in the encoding of the profiles. Why do you care that it isn't intuitive? How many people can lay their hands on a CIE reference light source? Then, having got your light source, now what? Why on earth do you need to? Are you manufacturing printing, scanning, or digital cameras? Only the manufacturers who are profiling their devices have to deal with it, and it is highly technical. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684
Re: filmscanners: Fw: Color Profiles for Scanners
On Sun, 21 Jan 2001 17:01:16 - Alan Tyson ([EMAIL PROTECTED]) wrote: And we should also, perhaps, remember that different persons' colour perceptions (Mk1 eyeball + brain software) may differ Absolutely. My right eye is about +2CC yellow compared to my right, or maybe it's the left which is about 2CC blue. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Frank writes ... As Frank implies ... there is a separate Lab defined for each illumination ... e.g., subjects illuminated under D65 vs D50. Not a big difference, Once the eye becomes accommodated, it shouldn't notice a difference between the two lightings, but to see that in fact there is a considerable difference, try switching your monitor between these two temperatures. You'll see a pretty distinct difference. The "small" difference I was referring to is ... converting from one space to another, D65 v D50, is I believe a single variable in a transform function. But all this is an aside. I didn't quite understand the points you were trying to make in your latest post, ... The important points (other than humbled by a blunder) were: 1st point ... relative to Photoshop color spaces, I now understand RGB is relative, and it is possible to create a profile for which you may find 0-0-255 in your gamut. 2nd point ... one should never hope to find 0-0-255 in your gamut because it would imply you have very little room to edit your "blue" component. sidelight point ... if you never want 0-0-255 in your gamut because of lack of editing overhead, then as you approach perfect devices and the wide gamuts associated with them, then 0-0-255 should approach a "perfect" blue ... a "concept" blue ... a "nonexistent" blue. Last point, really a question ... the blues our eye perceive have somehow been translated into computer a (mathematical) transform. That is, someone extracted "blue" from the spectrum of photons, and defined a physical reference point and a transform for computer display. I wonder if it isn't the CIELAB (L*a*b) definition ... that is, other Lab spaces seem (?) to be relative to L*a*b. Next question ... can we bring this subject back to film scanners, and their ability to capture the gamut of film (my golden fleece). shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
On Sat, 20 Jan 2001 20:18:41 -0800 shAf ([EMAIL PROTECTED]) wrote: What color definition allows us to equate what PS presents for us with nature?? I refer the honourable gentleman to the answer I gave earlier : the Mk1 eyeball. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
On Fri, 19 Jan 2001 22:28:59 + Photoscientia ([EMAIL PROTECTED]) wrote: CIE L*a*b* is poorly suited to DTP applications in three main areas: The main objection to CIELAB for DTP is (AFAIK) that DTP is (professionally) always done in CMYK for pre-press output - eg Pantone is CMYK gamut - and CIELAB is an RGB space. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Tony writes ... On Fri, 19 Jan 2001 22:28:59 + Photoscientia ([EMAIL PROTECTED]) wrote: CIE L*a*b* is poorly suited to DTP applications in three main areas: The main objection to CIELAB for DTP is (AFAIK) that DTP is (professionally) always done in CMYK for pre-press output - eg Pantone is CMYK gamut - and CIELAB is an RGB space. I would object to L*a*b, as any application space, simply because it is unintuitive. My own questions regarding it are only in the context of it being a absolute reference ... otherwise (practically) useless. shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
And we should also, perhaps, remember that different persons' colour perceptions (Mk1 eyeball + brain software) may differ. For example, my own blue sensitivity or perception clearly differs from the rest of family, because they are wont to say, on nice sunny days, "look at that beautiful blue lake" and I say "No, it's black, or nearly black". Photos of the scene then show an annoyingly bright blue lake. On these occasions I've seen what is about (0,0,255) according to everyone else as nearly black. There are also arguments over whether a particular turquoise sea colour is more green or more blue. (I pass all the colour blindness tests, BTW. Maybe I have polarising eyeballs.) Surprisingly, I detect no difference in taste over colour balance on monitors prints. I also notice that some people describe sodium vapour lights, (and the associated spectral lines in a spectroscope), as *orange* and some describe them as *yellow*. So even if a rigorous calibration system from original scene to monitor or print is possible, we still won't necessarily agree it looks right. This is just as well, as in all artistic subjects, where variety is everything. Regards, Alan T - Original Message - From: Tony Sleep [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Sunday, January 21, 2001 1:17 PM Subject: Re: filmscanners: Fw: Color Profiles for Scanners I refer the honourable gentleman to the answer I gave earlier : the Mk1 eyeball.
Re: filmscanners: Fw: Color Profiles for Scanners
I don't want to knock this thread. I've enjoyed the discussion, even though some of it started out to be over my head. However, for most of us, the ultimate product is usually a print that we can show to others, and have them accept that the colors make sense to them, as being representative of 'reality' (where that is the artistic objective). The observers are not likely to care much whether the blue is a perfect 0-0-255. If it is the sky, then it should elicit the response that it is what a sky looks like... Oil Painters will take a couple (or several) tubes and mix up a color that 'looks good' to the artist. Good painters are not ignorant of color theory, but I doubt they would get involved in a discussion like this thread. And, in general, an original oil painting is unique, and the artist would probably be hard put to duplicate it if asked to do so. Hersch At 10:05 PM 01/19/2001 +0100, you wrote: shAf wrote: I'd love to learn more from discussing this subject with my peers ... I need some understanding as much as anyone else, but I do believe my observations and questions are valid, and that I'll be able to contribute as well as learn. Absolutely the same for me ! Thanks to all the contributors ... reading and understanding is learning ... for me . Sincerely. Ezio www.lucenti.com e-photography site - Original Message - From: "shAf" [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, January 19, 2001 7:49 PM Subject: RE: filmscanners: Fw: Color Profiles for Scanners Tony writes ... I think this debate does belong here. Very few people, including me, understand all this stuff fully, yet it inescapably goes with the territory of film scanning. OK then ... but let's back up a bit, and agree on some concepts. Color exists, and devices cannot capture all of it. For example, we might begin with ignoring the quality of our lenses, and skip to something we have day-to-day control over, and realize that certain films are more-or-less sensitive to certain wavelengths. Do we want to understand this as the film's gamut?? (Andrew feel free to jump in ... please!). At this point in the discussion, I am at a loss to define the fixed point around which all other definitions of color (color spaces, device gamuts) are relative. I would love to believe this is the CIELAB color space, but I've read there are different versions of Lab color space. This is aggravating to me ... we need to first find that "fixed" reference point. It also seems to me RGB must somehow be "fixed", but there exists an anchor by which we define PS color spaces. From here we'll be able to understand to what degree film can capture nature ... a scanner can capture film ... the appropriatness of our Photoshop working spaces. I entered into this thread only to express an observation regarding understanding RGB color space and the gamut associated with it (assume any color space, but it began with the color spaces associated with Photoshop ... e.g., AdobeRGB). My point was one of curiosity ... not only are there colors (as defined by a RGB value) which are outside the working space, there also exists colors as defined by RGB values which do not exist at all. A question to ask here would be if anyone believes, that when they defined the RGB editing color space, if they didn't define it as such that the endpoints (the "pure" R,G B values) would never be actually found. That is, don't we define these editing color spaces to ^enclose^ anything we may encounter??? I'd love to learn more from discussing this subject with my peers ... I need some understanding as much as anyone else, but I do believe my observations and questions are valid, and that I'll be able to contribute as well as learn. Comments in the context of your post follow ... On Thu, 18 Jan 2001 20:42:09 -0800 shAf ([EMAIL PROTECTED]) wrote: Some values are so useless, they are even outside Lab ... and not only can you not bringthem into gamut, they should even be there in the first place ... especially if we're talking about photography. Right, now I understand better what you are getting at and have no argument with most of it (apart from the philosophical one that RGB is device dependent so the purity and intensity of an eg pure blue depends on the device, not the RGB value - I'll agree a device is capable of only some colors. With regard to monitors, there is much of the working space's gamut we simply have to accept on faith. We should be happy Photoshop's "monitor compensation" at least removes the display's bias (influence) on the perceived colors vs the RGB values. There is of course usefulness in out-of-gamut colors ... for, as you say, headroom, and because they can be brought into gamut. But surely you'll agree some RG
Re: filmscanners: Fw: Color Profiles for Scanners
"Hersch Nitikman" [EMAIL PROTECTED] wrote: perfect 0-0-255. If it is the sky, then it should elicit the response that it is what a sky looks like... Ultimately, if we're able to scan and print the pictures and we like the results, that's what really matters. :) in general, an original oil painting is unique, and the artist would probably be hard put to duplicate it if asked to do so. Not just oils. I know my mother had tried to make duplicates of some of her paintings, but they are seldom good enough for her satisfaction. Rob
RE: filmscanners: Fw: Color Profiles for Scanners
On Fri, 19 Jan 2001 10:49:53 -0800 shAf ([EMAIL PROTECTED]) wrote: Color exists, and devices cannot capture all of it. For example, we might begin with ignoring the quality of our lenses, and skip to something we have day-to-day control over, and realize that certain films are more-or-less sensitive to certain wavelengths. Do we want to understand this as the film's gamut?? (Andrew feel free to jump in ... please!). Unfortunately I just spotted in the listserver logs that Andrew has unsubscribed. fixed point around which all other definitions of color (color spaces, device gamuts) are relative. The colourspace of your vision is that fixed point, nothing else. CIELAB describes all the colours available in additive colour systems like computer screens. The reason it's useful for us is that it maps the eye's non-linear sensitivity to HSL. It's a working assumption about most peoples' vision, barring colour blindness or individual divergence from the norm. When you say 'different versions of CIELAB', maybe you are thinking of CIELUV, a different space which AFAICR does likewise for all the colours we can see in a subtractive system, ie where the illuminant is white light. I think CIELUV is supposed to describe our perception of the real world, where what our eyes get is what has not been subtractively absorbed by reflection. I'm guessing now, but perhaps the foregoing argument about real vs. non-real colours lays with CIELUV and CIELAB not overlapping perfectly. CIELAB is the wider space, I believe. Which means not that we cannot see some colours but that there are colours available within an additive system which don't occur subtractively. AFAICS all RGB spaces should fit within CIELAB as there is little merit in making output devices which work outside our visual gamut. It could be done, eg printing in infra-red reflective inks, but we'd have a hard time exhibiting them ;) If we are dealing with film, CIELAB is the relevant description because it's an additive system, and CIELAB describes how we see the film image. Each film imposes its own gamut. What we get on film is a remapping of scene values from the original colours to the colours and densities of which the film is capable. This is relative and 'device dependent' - the emulsion sensitivities and dye intensities comprise the 'device'. So far the process has been analogue, uncomplicated by RGB values or profiles - but it is the same qualitatively. The film's 'device characteristics' can be described as a profile. Then we scan it. The scanner has its own foibles and non-linearities in the manner in which it attempts to translate dye colours and densities to RGB values. We are now in the realms of colour managed workflow, requiring an input profile (film colorspace description), a working space profile (scanner colourspace description), and output profile (target colourspace description, eg Adobe 1998 or sRGB). Only if all three are available and accurate can we move from film image to RGB values in a file which are the most accurate possible mapping of the original scene. To view the file as accurately as possible, we require an input profile (description of the colourspace to which the RGB values relate, which was the output profile above, Adobe 1998 or whatever), a working space profile (profile of the software's RGB space - in PS, Adobe RGB), and an output profile (the monitor profile, a description of the monitor's working space). If all the above works properly, we get on screen, within the limitations of the monitor, the best possible representation of what the original scene looks like - minus the stuff irretrievably lost by the film, scanner, etc, and plus artifacts introduced by film grain, scanner defects such as noise and aliasing. It also seems to me RGB must somehow be "fixed", but there exists an anchor by which we define PS color spaces. Only by the process of successive profile-profile translations can RGB be linked back to CIELAB, or rather what we saw of the original scene. Any break in the chain - lack of, or inaccuracy in a profile - removes the automatic optimisation of the process. This is inherent with colour negative because film profiles are unavailable, and we have to rely on software tools and judgement to reconstitute a plausible set of RGB values. From here we'll be able to understand to what degree film can capture nature ... a scanner can capture film ... the appropriatness of our Photoshop working spaces. Choice depends wholly on what we intend doing with the image, how we intend outputting it. Different colourspaces relate better or worse to different output devices, we may lose more or less accuracy when we re-map to the output device's profiled space. ... not only are there colors (as defined by a RGB value) which are outside the working space AFAICS this really shouldn't be the case there also exists colors as defined by RGB
RE: filmscanners: Fw: Color Profiles for Scanners
On Fri, 19 Jan 2001 10:11:54 -0800 shAf ([EMAIL PROTECTED]) wrote: Which would imply RGB space is fixed by the display, and Adobe gamma provides relative compensation. Which would imply RGB space is fixed by the display, and Adobe gamma provides relative compensation. Of course, that is what Adobe gamma does. The intention is to ensure a given triplet of RGB values looks as similar as possible on any monitor - to compensate for device gamut variation as much as it can. But that doesn't mean RGB is in any sense device-independent. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
RE: filmscanners: Fw: Color Profiles for Scanners
On Fri, 19 Jan 2001 07:30:46 -0800 shAf ([EMAIL PROTECTED]) wrote: I am claiming you'll never be able to photograph an equivelent of RGB=255-0-0. If you do surely let the color community know :o) I'll avoid photographing my monitor display then. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Right on. Hersch At 11:06 PM 01/20/2001 +1000, you wrote: "Hersch Nitikman" [EMAIL PROTECTED] wrote: perfect 0-0-255. If it is the sky, then it should elicit the response that it is what a sky looks like... Ultimately, if we're able to scan and print the pictures and we like the results, that's what really matters. :) in general, an original oil painting is unique, and the artist would probably be hard put to duplicate it if asked to do so. Not just oils. I know my mother had tried to make duplicates of some of her paintings, but they are seldom good enough for her satisfaction. Rob
RE: filmscanners: Fw: Color Profiles for Scanners
On Fri, 19 Jan 2001 00:29 + (GMT) I ([EMAIL PROTECTED]) wrote: Since vanilla RGB is completely device dependent, you seem to be saying that the monitor's output gamut is a wider space than LAB? I wouldn't be surprised by this, and it would seem a valuable attribute. A monitor which only showed us a fraction of the image colours would be a pig to work with (try editing in 16 or 256colours:). Most of the colours in a high-bit file are 'imaginary' as they inhabit a wider range of values than the monitor is capable of, but this is a virtue we exploit when manipulating the image - the manipulation brings colours within a gamut that is useful and visible to us. So what? This last sentence is rubbish, of course. A high bit image in a given colour space occupies the exact same gamut as an 8bit file in the same colour space. Well it was about 3am. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
On Thu, 18 Jan 2001 20:42:09 -0800 shAf ([EMAIL PROTECTED]) wrote: Some values are so useless, they are even outside Lab ... and not only can you not bring them into gamut, they should even be there in the first place ... especially if we're talking about photography. Right, now I understand better what you are getting at and have no argument with most of it (apart from the philosophical one that RGB is device dependent so the purity and intensity of an eg pure blue depends on the device, not the RGB value - so whether you can see what you get depends on the device). But, even if in practice monitors can't realise R0 G0 B255, the point above doesn't seem correct anyway. It's advantageous to have a surfeit of colours in your working space, surely, as headroom for manipulating RGB values. Those manipulations might either bring unseeable and un-outputtable bit values back within a useable gamut, or force them out of it. Otherwise they would simply vanish. In all cases, these values will be later reprofiled to suit the output device (eg your monitor), and there would be something very wrong if the output profile didn't remap them to be within its capabilities. I think this debate does belong here. Very few people, including me, understand all this stuff fully, yet it inescapably goes with the territory of film scanning. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
RE: filmscanners: Fw: Color Profiles for Scanners
shAf wrote: I simply don't understand this esoteric point that 'all of the 16M colours don't exist'. I mean there is a color equivelent found in the natural world (anywhere, anything) for RGB=30-0-230, but not for 0-0-255 That's wrong. RGB is a relative system, and as such, 0,0,255 is what ever you assign to it, or calibrate it to.
RE: filmscanners: Fw: Color Profiles for Scanners
-Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of shAf Sent: Friday, January 19, 2001 7:31 AM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners shAf wrote: I am claiming you'll never be able to photograph an equivelent of RGB=255-0-0. If you do surely let the color community know :o) You still don't understand that RGB triads are not absolute colors, and so what you say does not make sense. They can only be *mapped* to absolute colors through some kind of color calibration process, and there's no way that you can claim that 255-0-0 does not map to a color in any particular space after I have calibrated my monitor, without looking at the color mapping in the profile file. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684
RE: filmscanners: Fw: Color Profiles for Scanners
Austin writes ... What color it is depends on what particular RGB device you are talking about: Exactly. RGB is relative. There is NO Pantone exact color chart for RGB. A fact which might challenge RGB being relative is that whenever you profile-to-profile, Photoshop changes the RGB values. This would imply it is the color spaces which are relative and are aligning to a fixed RGB definition. Another example, is that all other softwares simply accept RGB values (e.g., browsers), but what colors they do display are relative to display hardware and the software (drivers) associated. Which would imply RGB space is fixed by the display, and Adobe gamma provides relative compensation. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
I am claiming you'll never be able to photograph an equivelent of RGB=255-0-0. If you do surely let the color community know :o) You still don't understand that RGB triads are not absolute colors, and so what you say does not make sense. They can only be *mapped* to absolute colors through some kind of color calibration process, and there's no way that you can claim that 255-0-0 does not map to a color in any particular space after I have calibrated my monitor, without looking at the color mapping in the profile file. I do imagine what you describe being a possibility, but I can't imagine why. Don't imagine, take it as a fact. It was just done that way, because there was no requirement for super accurate color work when RGB monitors were first made. Red was just the red gun of the monitor...etc, and different manufacturers made different guns etc, so they are all relative, until they are calibrated.
Re: filmscanners: Fw: Color Profiles for Scanners
shAf wrote: I'd love to learn more from discussing this subject with my peers ... I need some understanding as much as anyone else, but I do believe my observations and questions are valid, and that I'll be able to contribute as well as learn. Absolutely the same for me ! Thanks to all the contributors ... reading and understanding is learning ... for me . Sincerely. Ezio www.lucenti.com e-photography site - Original Message - From: "shAf" [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Friday, January 19, 2001 7:49 PM Subject: RE: filmscanners: Fw: Color Profiles for Scanners Tony writes ... I think this debate does belong here. Very few people, including me, understand all this stuff fully, yet it inescapably goes with the territory of film scanning. OK then ... but let's back up a bit, and agree on some concepts. Color exists, and devices cannot capture all of it. For example, we might begin with ignoring the quality of our lenses, and skip to something we have day-to-day control over, and realize that certain films are more-or-less sensitive to certain wavelengths. Do we want to understand this as the film's gamut?? (Andrew feel free to jump in ... please!). At this point in the discussion, I am at a loss to define the fixed point around which all other definitions of color (color spaces, device gamuts) are relative. I would love to believe this is the CIELAB color space, but I've read there are different versions of Lab color space. This is aggravating to me ... we need to first find that "fixed" reference point. It also seems to me RGB must somehow be "fixed", but there exists an anchor by which we define PS color spaces. From here we'll be able to understand to what degree film can capture nature ... a scanner can capture film ... the appropriatness of our Photoshop working spaces. I entered into this thread only to express an observation regarding understanding RGB color space and the gamut associated with it (assume any color space, but it began with the color spaces associated with Photoshop ... e.g., AdobeRGB). My point was one of curiosity ... not only are there colors (as defined by a RGB value) which are outside the working space, there also exists colors as defined by RGB values which do not exist at all. A question to ask here would be if anyone believes, that when they defined the RGB editing color space, if they didn't define it as such that the endpoints (the "pure" R,G B values) would never be actually found. That is, don't we define these editing color spaces to ^enclose^ anything we may encounter??? I'd love to learn more from discussing this subject with my peers ... I need some understanding as much as anyone else, but I do believe my observations and questions are valid, and that I'll be able to contribute as well as learn. Comments in the context of your post follow ... On Thu, 18 Jan 2001 20:42:09 -0800 shAf ([EMAIL PROTECTED]) wrote: Some values are so useless, they are even outside Lab ... and not only can you not bringthem into gamut, they should even be there in the first place ... especially if we're talking about photography. Right, now I understand better what you are getting at and have no argument with most of it (apart from the philosophical one that RGB is device dependent so the purity and intensity of an eg pure blue depends on the device, not the RGB value - I'll agree a device is capable of only some colors. With regard to monitors, there is much of the working space's gamut we simply have to accept on faith. We should be happy Photoshop's "monitor compensation" at least removes the display's bias (influence) on the perceived colors vs the RGB values. There is of course usefulness in out-of-gamut colors ... for, as you say, headroom, and because they can be brought into gamut. But surely you'll agree some RGB values are so far out of gamut (relative to some device gamuts and small working spaces), they can't be considered useful (but surely archive the original image+gamut, you'll want them later) cheerios, shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
Hi Shaf, Frank, Austin, Tony et al. Sorry about jumping into this discussion with late replies to earlier points, but my e-mail's been on the fritz. shAf wrote: Frank writes ... And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Well, sorry Shaf, but my photometer can easily 'see' the difference between 0,0,255, and 0,0,254. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality. Oh come on. L*a*b* was made up long before high quality colour monitors existed (last revision 1978, I believe), and is a very esoteric and unintuitive 'standard.' A red-green, and a blue-yellow axis! How scientific or logical is that? Clark Maxwell must be spinning in his grave. L*a*b* doesn't define absolute colours anymore than RGB, the Lab values must be referred to a specific illuminating source, D65 for example. In reality, this means that Lab is biased toward reflective media, since there is no standard reference source for Lab with a colour temperature higher than 7500 Kelvin, and the white point of an RGB display cannot be given a true colour temperature. The observation angle of 2 or 10 degrees must be specified as well, which I admit I don't understand the significance of. So a set of L*a*b* figures mean nothing without a whole lot of other parameters being specified. Here is the conclusion of a very well researched and argued document, against the use of CIE L*a*b* for the purpose of DTP. "CONCLUSION CIE L*a*b* is poorly suited to DTP applications in three main areas: It correlates poorly with observed color differences. The cube root definition of L* does not model the perception of brightness in complex scenes; a square root law models human vision better. In the dark region (below 50% dot), CIE L*a*b* is a poor predictor of color errors in DTP. CIE L*a*b*, rather than giving us uniform color space, has opened color tolerancing up to question. It reports large errors where none exist, and it is not uniform in lightness or chroma. CIE L*u*v*, with a modified definition of L*, might provide a better basis for DTP. However, companion papers to this one show that Guth's ATD space, based on a full opponent model of human vision, is an even better choice. " For myself, I submit that the gamut of 'real' colours should now be extended to include super-saturated man-made sources as well. After all, the natural spectrum of the sun was only available in the diluted form of a rainbow, until Isaac Newton messed about with a glass prism behind the drawn curtains of his mum's best room. Scientifically, colour saturation only depends on how narrow and how relatively bright you can make the bandwidth of a chunk of the spectrum, or an admixture of spectral slices. If you must use the concept of 'unreal' colours, then anything magenta coloured between blue and red exists outside of the natural spectrum, and so cannot have a high saturation in nature, but strangely, can be defined by Lab numbers. I agree that RGB isn't any better at the moment, but that is what s(tandard)RGB is all about. It is at least an attempt to unify and standardise the colour displayed on monitors remote from each other, and has the backing of most major players in the computer display arena. OK. It has a smaller gamut than some other RGB spaces, but that is deliberately done in order to accomodate a wider variety of display devices. As long as manufacturers and users continue to ignore attempts to standardise, we will have the situation where 23,95,136 on my monitor looks nothing like 23,95,136 on yours. Just something to think about. Regards, Pete.
RE: filmscanners: Fw: Color Profiles for Scanners
Good grief. Say my green gun is half shot and is only putting out 1/2 it's specified power. Then 0,255,0 on my monitor will be smack in the middle of just about any color space you can name and monitor calibration software will map it there. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of shAf Sent: Friday, January 19, 2001 10:12 AM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners Austin writes ... What color it is depends on what particular RGB device you are talking about: Exactly. RGB is relative. There is NO Pantone exact color chart for RGB. A fact which might challenge RGB being relative is that whenever you profile-to-profile, Photoshop changes the RGB values. This would imply it is the color spaces which are relative and are aligning to a fixed RGB definition. Another example, is that all other softwares simply accept RGB values (e.g., browsers), but what colors they do display are relative to display hardware and the software (drivers) associated. Which would imply RGB space is fixed by the display, and Adobe gamma provides relative compensation. shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
"shAf" [EMAIL PROTECTED] wrote: Rob writes ... Can I just ask - do you mean they can't be reproduced by a reflective medium like a photograph? Surely pure red *does* exist in the natural world, even if it's only in the light of a rainbow or a laser beam for instance? I am claiming you'll never be able to photograph an equivelent of RGB=255-0-0. If you do surely let the color community know :o) OK, I can understand that it may not be possible to reproduce it accurately - *any* photography is a rough estimate of reality - but does the colour actually exist in *reality* not on film in the conditions I suggested? Surely the bright red in a spectrum from a prism, or a laser is pure red? Rob
RE: filmscanners: Fw: Color Profiles for Scanners
Triad? Look it up in the dictionary. It's not jargon. An absolute color specification would be any specific CIE XYZ tristimulus triad. Note, however, that CIE tristimulus values do not describe color appearance, because how a color appears depends on the ambience in which it is viewed and the viewer's state of adaptation. Tristimulus values are something objective, something that can be measured with a colorimeter. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of shAf Sent: Friday, January 19, 2001 10:24 AM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners shAf wrote: In the meantime, please define what you mean by "triad" and "absolute" ... examples of absolute colors please. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
shAf wrote: At this point in the discussion, I am at a loss to define the fixed point around which all other definitions of color (color spaces, device gamuts) are relative. I would love to believe this is the CIELAB color space, but I've read there are different versions of Lab color space. These are just different mappings of the original CIE tristimulus values so that they are spread out more evenly and correspond better to how humans perceive colors. The different versions are simply different operations on the original values, and different mappings have different advantages and disadvantages. This is aggravating to me ... Why on earth? Because human vision is nonlinear? Saying that you are aggravated by this amounts to being discontented with what hundreds of millions of years of evolution has done to our vision. Color science has simply uncovered the fact that our vision is not as straightforward as our simple minds would prefer. The world itself is more complex than evolution has prepared us for. Part of being a technological species is simply to learn to deal with it. we need to first find that "fixed" reference point. That's what scientific instruments like colorimeters are for: to measure what's really there, independent of our psychophysical and psychological perception. I entered into this thread only to express an observation regarding understanding RGB color space and the gamut associated with it And what you continue to fail to understand is that there *is* no gamut associated with RGB. There is a particular gamut presented by a particular monitor adjusted in a particular way, and how RGB maps into that gamut is going to be different for every single monitor, depending on what the gamut for that particular monitor actually is at any particular time. not only are there colors (as defined by a RGB value) which are outside the working space, there also exists colors as defined by RGB values which do not exist at all. Once again, colors are NOT defined by RGB. RGB is simply used to map the gamut of any particular monitor, which may produce a gamut radically different from the monitor sitting right next to it, yet they both process the same RGB numerical "space". RGB is simply a range of numbers that map over the particular gamut of a particular monitor. This is so obvious I can't figure out what your problem is. You can present 145-230-167 on three different uncalibrated monitors and you will see three different colors. There's absolutely nothing absolute about these numbers. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684
Re: filmscanners: Fw: Color Profiles for Scanners
Shaf wrote: Say you have a perfect camera, perfect film, and a perfect scanner ... and your image of a "natural" subject ends up in Photoshop. You will never see the pixel value 0-0-255 ... and in fact, there are a number of RGB values you'll never find. That's good. It wouldn't do to have colours out there that our digital darkrooms can't handle. Colin Maddock
RE: filmscanners: Fw: Color Profiles for Scanners
On Wed, 17 Jan 2001 15:47:39 -0800 shAf ([EMAIL PROTECTED]) wrote: It was then when I realized what RGB pixel values are ... 16 million possibilities, but only some of them actually are nature's real colors. I dare say a big part (but not most) of RGB is out of nature's gamut!!. Nature now includes monitor phosphors :) Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
RE: filmscanners: Fw: Color Profiles for Scanners
Duh! -Original Message- From: shAf [SMTP:[EMAIL PROTECTED]] Sent: Thursday, January 18, 2001 6:48 AM To: [EMAIL PROTECTED] Subject: Re: filmscanners: Fw: Color Profiles for Scanners [Oostrom, Jerry] [znibh..] Not much of a point really. I'm sure many readers are saying ... "duh". My original point was for someone somewhat befuddled with [Oostrom, Jerry] befuddled? I look zizzup 'n ze Vebsterz: befuddled- thorolly confused (wiff liquor) What makez u fink I'm drunk!?@#$% Hikh! [Oostrom, Jerry] [schnappss!] I just threw it in for conversation, not argument. shAf :o) Liquor_pfuh_Beer.jpg But anyway, thank you chef ! ;o) Liquor_pfuh_Beer.jpg
RE: filmscanners: Fw: Color Profiles for Scanners
Tony writes ... On Wed, 17 Jan 2001 10:02:16 -0800 shAf ([EMAIL PROTECTED]) wrote: ... without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut... which is the only color definition defined to come even close to physical reality. I think you are only illustrating that 16.4m colours far exceeds the discrimination of human vision, and that 8 bits should be enough for anything we intend looking at. Altho I agree 8bits is enough (barely) to provide human discrimination, I don't agree with any of the 16M colors being beyond human discrimination. Perhaps you can point to an example(?) I'm actually saying all of the 16M colors do not exist, in human perception or physical reality. You can, though, often see differences of ~3-5L if there is a suitable reference, though it depends on the colour and luminance. My "perfect" camera/film/scanner" analogy stated the point I wanted to make regarding RGB=0-0-255. Another would be a L*a*b conversion. That is, if you create an image of false colors ... 360 degree hue left-to-right and brightness vertically ... you'll have a representation of ALL (almost) RGB values. If you accept that Lab is the best model we have for human perception and the colors available in nature, then if you convert from RGB=Lab and then convert back (highbits please), you'll end up with an image which looks a lot like the original, but if you diff the original with the converted, you'll realize the number of pixels which fell outside Lab space. It is damn near half of the image! Of course anyone can point to some problems associated with my experiment ... conversion artifacts, the CM engine and its rendering intent ... but I've played with combinations, and all imply a similar number of RGB values which are outside Lab. Real colors are comprised of a volume inside the RGB "cube" ... RGB=0-0-255, and a number of other RGB values, are not inside the "real colors" volume". An interesting debate would be the acceptance of Lab color space as a representation of "real" colors. Why or why not is beyond my knowledge, but I've read a number of implications which would claim so. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
I've read from several different sources that our color perception can be trained and that some artists can perceived 10 million different colors. So 16 million colors does not *far* exceed our capabilities. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Tony Sleep Sent: Thursday, January 18, 2001 4:59 AM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners On Wed, 17 Jan 2001 10:02:16 -0800 shAf ([EMAIL PROTECTED]) wrote: How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality. I think you are only illustrating that 16.4m colours far exceeds the discrimination of human vision, and that 8 bits should be enough for anything we intend looking at. You can, though, often see differences of ~3-5L if there is a suitable reference, though it depends on the colour and luminance. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
RE: filmscanners: Fw: Color Profiles for Scanners
Nature now includes monitor phosphors :) Exactly. And even before they existed, you can't tell me it was impossible to find the colors they produce in "nature". I think this line of the conversation has been particularly unfruitful. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Tony Sleep Sent: Thursday, January 18, 2001 4:59 AM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners On Wed, 17 Jan 2001 15:47:39 -0800 shAf ([EMAIL PROTECTED]) wrote: It was then when I realized what RGB pixel values are ... 16 million possibilities, but only some of them actually are nature's real colors. I dare say a big part (but not most) of RGB is out of nature's gamut!!. Nature now includes monitor phosphors :) Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Tony writes ... A gamut comprises a subset of colours out of infinite variety. There are colours outside it which are simply unavailable and cannot appear in any image which uses it. ... Just to add ... the color gamut which is described by R,G B pixel values, whether it be wide gamut or 16bits/channel, is a manmade concoction of color definitions. There are a number of colors described by RGB which aren't even real ... for example 0,0,255. Nothing conceptualizes the definition of "gamut" better than "some colors described by RGB are outside reality's gamut". shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
Frank writes ... Relative to a monitor, 0,0,255 in itself is not a specific color, real or not. It depends on what the phosphors do when you feed it those values, ... That goes without saying ... And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality. shAf :o) Tony writes ... A gamut comprises a subset of colours out of infinite variety. There are colours outside it which are simply unavailable and cannot appear in any image which uses it. ... Just to add ... the color gamut which is described by R,G B pixel values, whether it be wide gamut or 16bits/channel, is a manmade concoction of color definitions. There are a number of colors described by RGB which aren't even real ... for example 0,0,255. Nothing conceptualizes the definition of "gamut" better than "some colors described by RGB are outside reality's gamut". shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
Is "physical reality" a technical term? And aren't you confusing color resolution with whatever "physical reality" is? If I grant that the phosphor response of monitors is flat between 0,0,240 and 0,0,255, how does this impinge on whether that color "real?" You are conflating two issues, so I can't follow what argument you are really trying to make. shAf writes ... Frank writes: And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality.
RE: filmscanners: Fw: Color Profiles for Scanners
Daniel writes ... Is "physical reality" a technical term? And aren't you confusing color resolution with whatever "physical reality" is? If I grant that the phosphor response of monitors is flat between 0,0,240 and 0,0,255, how does this impinge on whether that color "real?" You are conflating two issues, so I can't follow what argument you are really trying to make. I wasn't arguing anything ... merely stating a curiousity that is also fact ... rather a "believe it, or not" type of curiosity, which I thought, in the context of the original question was instructive. When Bruce Fraser explained to me (during a conversation about L*a*b) that many RGB "values" (e.g., 0-0-255) are manifestations of RGB which have no real counterpart in nature, it triggered an "a-ha". It was then when I realized what RGB pixel values are ... 16 million possibilities, but only some of them actually are nature's real colors. I dare say a big part (but not most) of RGB is out of nature's gamut!!. My own realization is that "RGB" is a human definition superimposed on nature. It is much like the limitations of our own language, when trying to make an abstract point (... is it true the French do it better?? ...) shAf :o) shAf writes ... Frank writes: And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality.
RE: filmscanners: Fw: Color Profiles for Scanners
I guess now the question is, what do you mean by "nature"? Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of shAf Sent: Wednesday, January 17, 2001 3:48 PM To: [EMAIL PROTECTED] Subject: RE: filmscanners: Fw: Color Profiles for Scanners Daniel writes ... Is "physical reality" a technical term? And aren't you confusing color resolution with whatever "physical reality" is? If I grant that the phosphor response of monitors is flat between 0,0,240 and 0,0,255, how does this impinge on whether that color "real?" You are conflating two issues, so I can't follow what argument you are really trying to make. I wasn't arguing anything ... merely stating a curiousity that is also fact ... rather a "believe it, or not" type of curiosity, which I thought, in the context of the original question was instructive. When Bruce Fraser explained to me (during a conversation about L*a*b) that many RGB "values" (e.g., 0-0-255) are manifestations of RGB which have no real counterpart in nature, it triggered an "a-ha". It was then when I realized what RGB pixel values are ... 16 million possibilities, but only some of them actually are nature's real colors. I dare say a big part (but not most) of RGB is out of nature's gamut!!. My own realization is that "RGB" is a human definition superimposed on nature. It is much like the limitations of our own language, when trying to make an abstract point (... is it true the French do it better?? ...) shAf :o) shAf writes ... Frank writes: And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality.
RE: filmscanners: Fw: Color Profiles for Scanners
Exactly. I don't get his point either. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of Daniel Weise Sent: Wednesday, January 17, 2001 1:53 PM To: '[EMAIL PROTECTED]' Subject: RE: filmscanners: Fw: Color Profiles for Scanners Is "physical reality" a technical term? And aren't you confusing color resolution with whatever "physical reality" is? If I grant that the phosphor response of monitors is flat between 0,0,240 and 0,0,255, how does this impinge on whether that color "real?" You are conflating two issues, so I can't follow what argument you are really trying to make. shAf writes ... Frank writes: And on my monitor, it DOES produce a real color, because I can SEE it. ... How can you say you "see" 0,0,255 when 0,0,254 is the same color?? ... I doubt you can start "seeing" any difference between these "pure" blues until 0,0,240 ... they are all the same ... especially in monitor space ... put up a gradient and prove it to yourself. Even without regard to monitor gamut, 0-0-255 falls outside the L*a*b gamut ... which is the only color definition defined to come even close to physical reality.
Re: filmscanners: Fw: Color Profiles for Scanners
Frank writes ... I guess now the question is, what do you mean by "nature"? Say you have a perfect camera, perfect film, and a perfect scanner ... and your image of a "natural" subject ends up in Photoshop. You will never see the pixel value 0-0-255 ... and in fact, there are a number of RGB values you'll never find. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
That sounds more like the limitations of a mapping algorithm than any limitations that we might find in nature. Frank Paris [EMAIL PROTECTED] http://albums.photopoint.com/j/AlbumList?u=62684 -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of shAf Sent: Wednesday, January 17, 2001 6:21 PM To: [EMAIL PROTECTED] Subject: Re: filmscanners: Fw: Color Profiles for Scanners Frank writes ... I guess now the question is, what do you mean by "nature"? Say you have a perfect camera, perfect film, and a perfect scanner ... and your image of a "natural" subject ends up in Photoshop. You will never see the pixel value 0-0-255 ... and in fact, there are a number of RGB values you'll never find. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
On Wed, 17 Jan 2001 10:37:58 +0100 Oostrom, Jerry ([EMAIL PROTECTED]) wrote: Now I just have to see if I can also see the limitations of gamut, of which you say they are much more apparent than granularity differences between color spaces used in 24/48 bit files. I have already seen some sort of posterization occur in the sky with editing certain high-bit scans of negative film in ProPhotoRGB where the scene was a heavy backlit one, with a tower in front and a bride in the shadow of that tower. Of course this could very well be caused by other factors as e.g. limitations of my scanwit. 'fraid so. For all the theory, hardware will impose its own limitations. What might be millions of colours can turn into a fraction of that, and as soon as you start pulling 'em around in software anything can happen. It's no criticism of the Acer, I've done extreme things to high bit files from several scanners and sooner or later even 16bits ain't enough. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
on 1/16/01 3:02 AM, Oostrom, Jerry at [EMAIL PROTECTED] wrote: *Is it correct to state that a color space with accompanying limits to its gamut will have only a finite number of colors? Yup. Will a smaller gamut color space allow finer granularity to code colors in its gamut than what a larger gamut color space would allow for the same range of colors Sort of. One of the "problems" with really wide gamut spaces is that the bits are spread farther apart and thus are not really appropriate for editing in 8 bits per color. A user who is editing in Wide Gamut RGB would really want a high bit file or you could end up with either banding or posterization. You should check out Bruce Fraser's fine article on the subject at: http://www.creativepro.com/story/feature/8582.html Andrew Rodney
Re: filmscanners: Fw: Color Profiles for Scanners
In a message dated 1/11/2001 2:12:10 PM EST, [EMAIL PROTECTED] writes: Your question would beg another ... "Is your scanner capable of a larger gamut than sRGB?" If not, then your PS working color space may as well be sRGB, but you don't lose anything if the scanner embeds sRGB and you subsequently convert to AdobeRGB when you open the file That's what I thought, too. Maybe Ed knows what the firmware allows. Most scanners return raw data straight from the CCD. Some scanners do color conversion internally. The Epson scanners all (optionally) convert colors to sRGB before returning it to the host computer (this is the mode that VueScan uses). Other scanners (i.e. HP) let you download a 3x3 matrix to do the color transform in the scanner, but I never use this in VueScan. Regards, Ed Hamrick
Re: filmscanners: Fw: Color Profiles for Scanners
Interesting questions indeed. I wonder what ''Wizard Ed'' is ''feeling'' about them. P.S. : Wizard ED is coming (nickname) to WizEd ! ;-) Sincerely. Ezio www.lucenti.com e-photography site - Original Message - From: "Tony Sleep" [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Thursday, January 11, 2001 7:58 PM Subject: RE: filmscanners: Fw: Color Profiles for Scanners On Wed, 10 Jan 2001 07:47:44 -0800 shAf ([EMAIL PROTECTED]) wrote: Your question would beg another ... "Is your scanner capable of a larger gamut than sRGB?" If not, then your PS working color space may as well be sRGB, but you don't lose anything if the scanner embeds sRGB and you subsequently convert to AdobeRGB when you open the file That's what I thought, too. Maybe Ed knows what the firmware allows. testing. (... which brings up a personal peeve ... "Why don't scanner test/reviews give us some idea of the color capacity of scanners?" Noted, with some dismay :) All you get out are RGB values within whatever space, too many variables including the user's settings of software for objective measurements AFAICS. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Bob writes ... I would think you might gain something if you perform tonal or color editing in PS: Might not results of the editing operation expand into the larger AdobeRGB gamut? In theory yes ... but the addition gamut would be beyond your display, and you wouldn't be able to see what you're doing(?) But if changes do expand into the larger gamut it might affect printed output. If you're not "able to see what you're doing" in a larger gamut then you would equally not see what you're doing when editing a file that came in a larger color space from a scanner. So that limitation, if significant, wound not seem to influence choice of editing color space. The choice is made more significant with PS6 ... it provides a few tools for editting in a larger gamut (e.g., soft proofing and display saturation control). My response was with respect to why use a larger gamut when editting an image from a small gamut device. For example ... a real example. Say you have two images of a freshly painted blue door, from a small gamut device and a large. Also say, the sun shining on the fresh gloss extends detail only into areas the large gamut device can capture. While such detail will not be evident on your display it exists none-the-less, but not at all in the small gamut image. A knowledgeable PS user would be aware of this, and appreciative of the PS6 tools. Now assume the desired working color space is a medium gamut AdobeRGB. If you expand the small gamut into medium it will not create detail ... however, if you squash large into medium the conversion will try to keep the detail (if rendering intent is "perceptual" rather than "relative colorimetric"). The point is "detail", and the question is it present and editable ... in one case yes, the other no. Getting back to the sRGB device and the utility of using AdobeRGB instead. It is a difficult question. As editting spaces, both are dependable, but a user will probably find more peers using AdobeRGB. With regard to printers it depends on which respond better, but AdobeRGB does extend better into cyans and yellows ... you would think AdobeRGB would be better ... and I'll not ever argue with anyone who converts a sRGB device space to AdobeRGB because it seems to work best (even if the desired detail not existing in the original gamut did not transfer). I sometimes wonder if film-scanner users are aware of large gamuts in film ... but most are. My own feeling is it is the detail which makes most photographs interesting, and that detail isn't only a matter for 2700ppi vs 4000ppi ... much of the detail is in the color! shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
Robert writes ... regarding my response ... \ I think you better examine the RGB pixel values before and after a profile-to-profile ... shAf :o) I think... Profile to profile changes the file(pixel values) but changing the "so called colour space" or you working space should not. When it is saved it will have the new profile though. If you ues the color sampler to select samples in the image, and apply "profile to profile" you will see changed pixel values. If you change your working space you will not see changes. You are correct ... I was thinking this was in the context of "converting" a sRGB profile to AdobeRGB. Changing the "working space" does not change the RGB values, but does change their appearence. shAf :o)
Re: filmscanners: Fw: Color Profiles for Scanners
on 1/10/01 7:54 PM, Bob Shomler at [EMAIL PROTECTED] wrote: But if changes do expand into the larger gamut it might affect printed output. Taking a file in sRGB and converting it to Adobe RGB isn't going to expand the gamut of the file. It's fixed after becoming sRGB. You can't increase the color gamut simply by converting into a space that can hold a larger number of colors. Andrew Rodney
Re: filmscanners: Fw: Color Profiles for Scanners
Taking a file in sRGB and converting it to Adobe RGB isn't going to expand the gamut of the file. It's fixed after becoming sRGB. You can't increase the color gamut simply by converting into a space that can hold a larger number of colors. Andrew Rodney Andrew, I'm curious if this will hold true if one performs color (and maybe tonal) editing operations on the file in the larger editing color space -- could that editing cause color to extend into areas present only in the larger space? More particularly, if one opens a sRGB file into PS with sRGB work space, edits colors and tones and prints image on a good color printer; then opens the same file into PS with a larger-than-sRGB gamut space such as Adobe RGB converting file on open, performs the same color and tonal edits and prints the edited file: might the printed images appear different?
Re: filmscanners: Fw: Color Profiles for Scanners
On Wed, 10 Jan 2001 23:17:28 + photoscientia ([EMAIL PROTECTED]) wrote: What, like a larger gamut than pixel levels from 0 to 255? g Um, well, AIUI eg R3 G192 B252 is not the same colour in AdobeRGB as it is in sRGB, RGB values change if converted. Nor should it display or print the same, via software which understands profiles. PS will be mapping those values from within the tagged gamut to the calibrated monitor profile, which then turns 'em into screen colour/luminance values within the gamut of the device. So what you see on screen should have a proportionality (can't think of a better word) to the tagged gamut, which will be different depending on which gamut it is. I can see no difference between files open side-by-side in PSP (sRGB), and PS5.5 (Adobe RGB). A mode-to-mode conversion from Adobe RGB to sRGB, or vice versa, takes almost zero time and makes NO visual difference to the image on screen. If I convert a scan made with Vuescan+Polaroid 4000 from Colourmatch RGB to sRGB, or any other profile, there's a clear change in histogram, displayed values and out of gamut warning. No visual difference suggests to me that the gamut of the scanner is hardwired to sRGB, so converting to a wider space achieves nothing extra. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
RE: filmscanners: Fw: Color Profiles for Scanners
On Wed, 10 Jan 2001 07:47:44 -0800 shAf ([EMAIL PROTECTED]) wrote: Your question would beg another ... "Is your scanner capable of a larger gamut than sRGB?" If not, then your PS working color space may as well be sRGB, but you don't lose anything if the scanner embeds sRGB and you subsequently convert to AdobeRGB when you open the file That's what I thought, too. Maybe Ed knows what the firmware allows. testing. (... which brings up a personal peeve ... "Why don't scanner test/reviews give us some idea of the color capacity of scanners?" Noted, with some dismay :) All you get out are RGB values within whatever space, too many variables including the user's settings of software for objective measurements AFAICS. Regards Tony Sleep http://www.halftone.co.uk - Online portfolio exhibit; + film scanner info comparisons
Re: filmscanners: Fw: Color Profiles for Scanners
Hi Rob, Shaf. I think you better examine the RGB pixel values before and after a profile-to-profile ... shAf :o) I think... Profile to profile changes the file(pixel values) but changing the "so called colour space" or you working space should not. When it is saved it will have the new profile though. If you ues the color sampler to select samples in the image, and apply "profile to profile" you will see changed pixel values. If you change your working space you will not see changes. Thanks Rob, that was the source of confusion, and why I saw no visible change. I stand corrected Shaf. I find that Photoshop is quite capable of buggering up a perfectly good file at a single mouse click. (Which is why I haven't experimented much with profile-to-profile) Most of the profile changes simply seem to be gamma changes or simple hue shifts, and I can see the point of those, apart from "wide gamut RGB" which seems to do the equivalent of desaturating the image colour by 40%. Excuse what might be a daft question, but what device exists that displays such a file correctly? And why would anyone want to create files that can only be displayed on such a device? Also, how come the existing white point stays constant if the colour space is supposedly expanded? Surely white should be 'desaturated' too, down to grey. Sorry, but the only way that I can see of changing my 'real' colour space, is to waggle the contrast on my monitor up and down, and/or change the type of paper and ink in my printer. Regards, Pete.
Re: filmscanners: Fw: Color Profiles for Scanners
on 1/11/01 5:21 PM, photoscientia at [EMAIL PROTECTED] wrote: I find that Photoshop is quite capable of buggering up a perfectly good file at a single mouse click. It's usually Photoshop users, not Photoshop that does this! Most of the profile changes simply seem to be gamma changes or simple hue shifts, and I can see the point of those, apart from "wide gamut RGB" which seems to do the equivalent of desaturating the image colour by 40%. The image isn't desaturated, it just looks that way when you view a file that isn't in Wide Gamut RGB with that space loaded. Excuse what might be a daft question, but what device exists that displays such a file correctly? Wide Gamut RGB? Not a one. In fact there are colors in Adobe RGB 1998 that fall outside monitor gamut. But just because you can't see it on screen doesn't mean you can't print it out. It just makes editing dicey if you make color moves you can't see. That's pretty rare but possible. The alternative is to keep your files in a narrow space that fully falls into your display gamut. Then your output can suffer. If you don't mind getting about 85% pure cyan on a CMYK press (itself not the widest gamut printer around by a long shot) then stick with sRGB. Andrew Rodney
RE: filmscanners: Fw: Color Profiles for Scanners
Rob writes ... I have an Acer Scanwit and under the scanner properties in Win98se it lists an sRGB color profile as the only profile associated with the scanner. My colorspace for Photoshop is AdobeRGB. Do I need associate AdobeRGb with the scanner or when I set AdobeRGB as my colorspace in Vuescan will the scan come into photoshop with AdobeRGB embedded? Your question would beg another ... "Is your scanner capable of a larger gamut than sRGB?" If not, then your PS working color space may as well be sRGB, but you don't lose anything if the scanner embeds sRGB and you subsequently convert to AdobeRGB when you open the file (... but you don't gain anything either ...). I suspect your scanner can deliver a better gamut than sRGB, and you are better off if you associate your Vuescan with your preferred working color space (AdobeRGB) ... but my suspicions would need be confirmed with some testing. (... which brings up a personal peeve ... "Why don't scanner test/reviews give us some idea of the color capacity of scanners?" ...) I could not find an .icm file called AdobeRBG, but since Pshop's using AdobeRGB I guess it has to be on my system somewhere, ... It would depend on your OS ... if Win98 it is in your 'windows/system/color' directory ... if Win2000, it is in 'winNT/system32/spool/color/' directory (... I haven't a clue about Macs ... in your 'colorsync' folder? ...) Or, since I am getting good prints with the setup as it is (color on the monitor matches pretty well with what I'm getting out of the printer) should I just leave it alone? Am I limiting the scanner input with this profile? Would another profile give me a wider range of colors? There is something to be said about "if it works, don't fool with it" ... but you probably are losing some control for editting your scanner's true color gamut if using sRGB is short changing it. shAf :o)
RE: filmscanners: Fw: Color Profiles for Scanners
Your question would beg another ... "Is your scanner capable of a larger gamut than sRGB?" If not, then your PS working color space may as well be sRGB, but you don't lose anything if the scanner embeds sRGB and you subsequently convert to AdobeRGB when you open the file (... but you don't gain anything either ...). I would think you might gain something if you perform tonal or color editing in PS: Might not results of the editing operation expand into the larger AdobeRGB gamut? -- Bob Shomler http://www.shomler.com/gallery.htm
Re: filmscanners: Fw: Color Profiles for Scanners
- Original Message - From: shAf [EMAIL PROTECTED] To: [EMAIL PROTECTED] Sent: Wednesday, January 10, 2001 6:31 PM Subject: RE: filmscanners: Fw: Color Profiles for Scanners Rob writes ... "photoscientia" [EMAIL PROTECTED] wrote: My understanding is that changing the so-called colour space in Photoshop simply changes the embedded profile, without actually making any difference to the image data itself. Well, that makes perfect sense - as I understand it, the profile is simply a mapping between numeric values and actual colours. I think you better examine the RGB pixel values before and after a profile-to-profile ... shAf :o) I think... If you change the "so called colour space" or your working space, you will not change the pixel values. If you do "profile to profile" you will. If you use the color sampler tool to select points in the image, you can confirm this by doing a working space change and then a profile to profile change. The original question was what to do with an untaged file from a scanner the outputs sRGB (as I recall). Simply convert it to your working space before editing. Bob Wright