Physics Talk Physics Talk    FAQ   Search   Memberlist   Usergroups   Register   Profile   Log in to check your private messages   Log in

Education Forums

Printing a Target by Absolute Colorimetry

Physics Talk Forum Index -> Color and Light

View previous topic :: View next topic
Author	Message
Gernot Hoffmann Guest	Posted: Fri Jun 15, 2007 8:03 pm    Post subject: Printing a Target by Absolute Colorimetry An IT8 target was programmed as an EPS file for sRGB and as an EPS file for CIELab. The files were packed as PDF-sRGB and PDF-Lab. Each PDF, opened in Photoshop in the respective color space, shows the correct numbers (rounded). The target should be printed with absolute colorimetry. The paper is blue-ish, therefore paper white should be corrected by yellow ink. I'm using the RIP ColorGate Productionserver with these choices for input profiles: RGB: ICC-profile / for Raster - for Vector / Intent CMYK: ICC-profile / for Raster - for Vector / Intent Gray: ICC-profile / for Raster - for Vector / Intent There is no choice for Lab inputs. The output profile is used as defined by profiling. a) Printing PDF-sRGB Rendering intent AbsCol is impossible, because sRGB is a D65 working space and reference white for Lab is D50. The print would be blue-ish. Rendering intent RelCol is OK. The measured values (relative to paper white) are not bad. White paper is (of course) not covered by yellow ink. b) Printing PDF-Lab Standard settings sRGB / ISO Coated / ISO Coated, each RelCol, deliver visually correct color patches, but the gray patches are greenish. Settings Default / Default / Default, each AbsCol, deliver a pretty good appearance with neutral grays, but the measured values are slightly wrong (worse than for PDF-sRGB), and white paper is not covered by yellow ink. Now the questions: 1) How should a RIP treat Lab files ? Who knows a RIP with Lab input definition (besides spot colors) ? 2) How should the target be coded for correct printing ? Perhaps as an RGB file for an RGB space with D50 white point, and then printed by AbsCol ? Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Sat Jun 16, 2007 11:08 am    Post subject: Re: Printing a Target by Absolute Colorimetry Dale Kelly schrieb: Quote: On Fri, 15 Jun 2007 08:03:24 -0700, Gernot Hoffmann wrote: There is no choice for Lab inputs. can you add profiles to the RIP? if so, obtain or create a Lab profile and add it to the RIP. how did you get your IT8 target into an EPS file? if you scanned it, you might just want to dispense with sRGB and CIELAB and just add your scanner's profile to the RIP, this would be much more accurate -- Dale http://www.vedantasite.org see website for real email address The targets were programmed by PostScript as EPS files, placed on pages and exported as PDFs. One set contains sRGB numbers (based on Lab), the other plain Lab numbers. According to Photoshop both PDFs are correct. I don't have the choice for using an 'identity Lab profile' as input. Input files are available for RGB, CMYK and Gray. Neverthless, there seems to be an automatism for interpreting Lab inputs more or less correct. Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Dale Kelly Guest	Posted: Sat Jun 16, 2007 11:08 am    Post subject: Re: Printing a Target by Absolute Colorimetry On Fri, 15 Jun 2007 08:03:24 -0700, Gernot Hoffmann wrote: Quote: There is no choice for Lab inputs. can you add profiles to the RIP? if so, obtain or create a Lab profile and add it to the RIP. how did you get your IT8 target into an EPS file? if you scanned it, you might just want to dispense with sRGB and CIELAB and just add your scanner's profile to the RIP, this would be much more accurate -- Dale http://www.vedantasite.org see website for real email address
Back to top
	Ads
Advertising Sponsor
Alastair M. Robinson Guest	Posted: Sat Jun 16, 2007 4:23 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Hi, Gernot Hoffmann wrote: Quote: 2) How should the target be coded for correct printing ? Perhaps as an RGB file for an RGB space with D50 white point, and then printed by AbsCol ? When I was experimenting with printer profiles and IT8.7 targets a while ago, this is how I did it. I fiddled with the numbers in your EPS file (which you kindly showed me some time ago) to imitate one of the many RGB spaces described on Bruce Lindbloom's site. Then rasterized it into an RGB TIFF. It would also be easy enough to create a new space, and synthesize a profile for it using LittleCMS. I printed the TIFF using an Abs. Col. transform between the RGB profile and the printer profile. The result was quite impressive (within the printer's gamut, obviously) - certainly a lot better than the results achieved by printing a scan of a target using a scanner->printer transform. Sadly, I can't find the files right now, otherwise I'd go into a bit more detail... All the best, -- Alastair M. Robinson
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Sat Jun 16, 2007 7:15 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Alastair M. Robinson schrieb: Quote: Hi, Gernot Hoffmann wrote: 2) How should the target be coded for correct printing ? Perhaps as an RGB file for an RGB space with D50 white point, and then printed by AbsCol ? When I was experimenting with printer profiles and IT8.7 targets a while ago, this is how I did it. I fiddled with the numbers in your EPS file (which you kindly showed me some time ago) to imitate one of the many RGB spaces described on Bruce Lindbloom's site. Then rasterized it into an RGB TIFF. It would also be easy enough to create a new space, and synthesize a profile for it using LittleCMS. I printed the TIFF using an Abs. Col. transform between the RGB profile and the printer profile. The result was quite impressive (within the printer's gamut, obviously) - certainly a lot better than the results achieved by printing a scan of a target using a scanner->printer transform. Sadly, I can't find the files right now, otherwise I'd go into a bit more detail... All the best, -- Alastair M. Robinson Alastair thanks for the feedback. This doc http://www.fho-emden.de/~hoffmann/a3gencolorhigh.pdf contains now on p.5 to p.7 three versions of the digital target: sRGB, ProPhotoRGB and Lab coding. ProPhotoRGB uses D50 white, therefore AbsCol will probably not make the print blue-ish (as it happens for sRGB). All versions are essentially vector graphics, small by file size. By the way: each page contains a target with numbers and another one without. Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Roger Breton Guest	Posted: Sat Jun 16, 2007 10:06 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Gernot, Quote: 1) How should a RIP treat Lab files ? I guess the same way Photoshop or argyll would. Quote:  Who knows a RIP with Lab input definition (besides spot colors) ? I think BEST colorproof. GMG Colorproof, but I'll have to check. In general, Gernot, you know this better than me, but *any* PostScript RIP has to accept Lab elements, otherwise they are not supporting the full PostScript specs. No? Quote: 2) How should the target be coded for correct printing ? For Lab source images and graphic elements? Quote: Perhaps as an RGB file for an RGB space with D50 white point, and then printed by AbsCol ? That's how I would do it, yes. Quote: Best regards --Gernot Hoffmann Roger Breton
Back to top
	Ads
Advertising Sponsor
Gerhard Fuernkranz Guest	Posted: Sat Jun 16, 2007 11:41 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Roger Breton wrote: Quote: In general, Gernot, you know this better than me, but *any* PostScript RIP has to accept Lab elements, otherwise they are not supporting the full PostScript specs. No? If I remember correctly, the PostScript language does NOT directly define a "Lab" color space. It is however indirectly possible to create a CIEbasedABC color space, such that its three color components represent L*, a* and b* (see the example given in the PLRM). The PDF on the other hand does directly define CIE-based "Lab" color spaces. And yes, it is basically my understanding too, that any PostScript RIP which claims to support at least PS Level 2, or any PDF RIP has to support CIE-based color spaces (and PS Level 3 adds a few additional types of CIE-based color spaces, which must be understood too by a RIP which claims to support PS level 3). Regards, Gerhard
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Sun Jun 17, 2007 11:02 am    Post subject: Re: Printing a Target by Absolute Colorimetry Gerhard Fuernkranz schrieb: Quote: Roger Breton wrote: In general, Gernot, you know this better than me, but *any* PostScript RIP has to accept Lab elements, otherwise they are not supporting the full PostScript specs. No? If I remember correctly, the PostScript language does NOT directly define a "Lab" color space. It is however indirectly possible to create a CIEbasedABC color space, such that its three color components represent L*, a* and b* (see the example given in the PLRM). The PDF on the other hand does directly define CIE-based "Lab" color spaces. And yes, it is basically my understanding too, that any PostScript RIP which claims to support at least PS Level 2, or any PDF RIP has to support CIE-based color spaces (and PS Level 3 adds a few additional types of CIE-based color spaces, which must be understood too by a RIP which claims to support PS level 3). Regards, Gerhard Gerhard and Roger, yes, PostScript allows to code colors by Lab. It's needs an additional procedure or dictionary. Such an EPS file is readable by the common Adobe programs. As well the PDF where the EPS was placed. In other words: the PS interpreters are OK. RIP ColorGate Productionserver accepts Lab, but the input parameters are nowhere defined. The result is, depending on RGB/CMYK/Gray settings, either almost correct or very wrong. I'll check this for Best Colorproof. But it's of no use, because the old version doesn't support CID fonts and the new version doesn't support my inkjet .. Instead of using the inkjet (office) I had printed the ProPhotoRGB version by a calibrated toner printer (at home). The result for AbsCol is very pleasing: neutral grays and indeed some yellow ink on plain paper areas, because the paper is blue-ish (Neusiedler Color Copy). Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Gerhard Fuernkranz Guest	Posted: Mon Jun 18, 2007 3:09 am    Post subject: Re: Printing a Target by Absolute Colorimetry Gernot Hoffmann wrote: Quote: Gerhard and Roger, yes, PostScript allows to code colors by Lab. It's needs an additional procedure or dictionary. Such an EPS file is readable by the common Adobe programs. As well the PDF where the EPS was placed. In other words: the PS interpreters are OK. RIP ColorGate Productionserver accepts Lab, but the input parameters are nowhere defined. The result is, depending on RGB/CMYK/Gray settings, either almost correct or very wrong. I'll check this for Best Colorproof. But it's of no use, because the old version doesn't support CID fonts and the new version doesn't support my inkjet .. Instead of using the inkjet (office) I had printed the ProPhotoRGB version by a calibrated toner printer (at home). The result for AbsCol is very pleasing: neutral grays and indeed some yellow ink on plain paper areas, because the paper is blue-ish (Neusiedler Color Copy). Gernot, I would indeed expect some ink on "unprinted" areas (which are not covered by any graphics objects in the document being printed), if the RIP/printer is operated in a proofing/emulation mode (i.e. if for instance the look of a press print is simulated, including the simulation of the paper color of e.g. SWOP or ISO coated). But I'm rather unsure whether I would really expect this behavior in "normal" print mode, where the RIP just drives the printer natively. I think I would rather expect that unprinted areas are actually left unprinted then. Note, which paper color should be simulated in normal print mode (when the RIP/printer does not emulate e.g. SWOP or ISO coated, etc.)? (Or does your PS or PDF document explicitly fill the whole page with a particular color (e.g. the IT8's Dmin color), before drawing the other objects of the IT8 target?) Regards Gerhard
Back to top
	Ads
Advertising Sponsor
Graeme Gill Guest	Posted: Mon Jun 18, 2007 6:03 am    Post subject: Re: Printing a Target by Absolute Colorimetry Gerhard Fuernkranz wrote: Quote: But I'm rather unsure whether I would really expect this behavior in "normal" print mode, where the RIP just drives the printer natively. I think I would rather expect that unprinted areas are actually left unprinted then. Note, which paper color should be simulated in normal print mode (when the RIP/printer does not emulate e.g. SWOP or ISO coated, etc.)? It depends on the RIP implementation. In my experience the authors of the RIP software typically do not think to run the paper white though the color transformation before clearing the page, leaving the page at output white, rather than the incoming colorspace white. This makes such a RIP unsuitable for proofing use. Making PostScript do an absolute transform using the PostScript machinery is possible but tricky, due to assumptions in the language that all color transformations are white relative. Graeme Gill.
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Mon Jun 18, 2007 9:04 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Gerhard Fuernkranz schrieb: Quote: Gernot, I would indeed expect some ink on "unprinted" areas (which are not covered by any graphics objects in the document being printed), if the RIP/printer is operated in a proofing/emulation mode (i.e. if for instance the look of a press print is simulated, including the simulation of the paper color of e.g. SWOP or ISO coated). But I'm rather unsure whether I would really expect this behavior in "normal" print mode, where the RIP just drives the printer natively. I think I would rather expect that unprinted areas are actually left unprinted then. Note, which paper color should be simulated in normal print mode (when the RIP/printer does not emulate e.g. SWOP or ISO coated, etc.)? (Or does your PS or PDF document explicitly fill the whole page with a particular color (e.g. the IT8's Dmin color), before drawing the other objects of the IT8 target?) Regards Gerhard Gerhard, good questions. The docs consist always of graphics on plain paper. The question is, whether blue-ish plain paper is covered by yellow ink for AbsCol. Unfortunately, the tests results are not simple: 1. Toner printer OKI 9600: yes 2. Inkjet Mutoh 6100 Onyx Postershop: No Best Colorproof XXL: No ColorGate Productionserver: No On the other hand I HAD two years ago yellow ink, but I don't remember the settings (Best ?). Three RIPs, three different sets of features: Productionserver has indeed a mode Production (posters) and a mode Proof. But in mode Proof only CMYK input profiles are accepted. Colorproof has indeed an input mode LAB , using an identity profile (this concerns the original question). Printing the mentioned target by inkjet delivers indeed very reasonable results, as measured for samples by DTP22, and by appearance. But plain paper isn't yellow-ish. A preliminary conclusion: the standardization should be improved. AbsCol concerns at present the relevance of paper color AND a strong shift from sRGB D65 to reference white D50. The latter effects prevents from applying AbsCol to D65 sources in this context. Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Gerhard Fuernkranz Guest	Posted: Tue Jun 19, 2007 2:21 am    Post subject: Re: Printing a Target by Absolute Colorimetry Gernot Hoffmann wrote: Quote: Gerhard, good questions. The docs consist always of graphics on plain paper. The question is, whether blue-ish plain paper is covered by yellow ink for AbsCol. Unfortunately, the tests results are not simple: 1. Toner printer OKI 9600: yes 2. Inkjet Mutoh 6100 Onyx Postershop: No Best Colorproof XXL: No ColorGate Productionserver: No On the other hand I HAD two years ago yellow ink, but I don't remember the settings (Best ?). Three RIPs, three different sets of features: Productionserver has indeed a mode Production (posters) and a mode Proof. But in mode Proof only CMYK input profiles are accepted. Colorproof has indeed an input mode LAB , using an identity profile (this concerns the original question). Gernot, I did reply to Graeme's message, but I don't see my follow-up message appearing in the newsgroup, and unfortunately I haven't kept a copy either :-( In this message I was saying that of course every object drawn by a PS/PDF document has an associated input colorspace, but I was still wondering which colorspace should be associated with the page background color, which is not associated with any object drawn by the document. And I think I found now the answer in the description of the "erasepage" operator in the PLRM. Note, "erasepage" is called by the PS interpreter internally at the start of each page. Quoted from the PLRM, page 581: "erases the current page by painting it with gray level 1.0 (which is ordinarily white, but may be some other color if an atypical transfer function has been defined). The entire page is erased, without reference to the clipping path currently in force. erasepage affects only the contents of raster memory; it does not modify the graphics state, nor does it cause a page to be transmitted to the output device. The showpage operator automatically invokes erasepage after imaging a page [...]" I think this implies, that the page background color is defined as a color in the DeviceGray colorspace, so I guess that any remapping of DeviceGray to a CIE-based colorspace via the DefaultGray resource is expected to influence the rendering of the page background color too. If I remember correctly, you said that you specified ISOcoated with absolute intent as "gray input profile", and I think now, that this particular setting is the cause for the yellow ink covering the blue-ish paper. Since absolute intent is used, this result in a simulation of the ISOcoated paper color. Regards, Gerhard Quote: Printing the mentioned target by inkjet delivers indeed very reasonable results, as measured for samples by DTP22, and by appearance. But plain paper isn't yellow-ish. A preliminary conclusion: the standardization should be improved. AbsCol concerns at present the relevance of paper color AND a strong shift from sRGB D65 to reference white D50. The latter effects prevents from applying AbsCol to D65 sources in this context. Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Wed Jun 20, 2007 11:19 am    Post subject: Re: Printing a Target by Absolute Colorimetry Quote: I think this implies, that the page background color is defined as a color in the DeviceGray colorspace, so I guess that any remapping of DeviceGray to a CIE-based colorspace via the DefaultGray resource is expected to influence the rendering of the page background color too. If I remember correctly, you said that you specified ISOcoated with absolute intent as "gray input profile", and I think now, that this particular setting is the cause for the yellow ink covering the blue-ish paper. Since absolute intent is used, this result in a simulation of the ISOcoated paper color. Gerhard, thanks for the suggestion, but I'm not able to reproduce the effect, which I had two years ago for proof printing: 'blue-ish paper gets yellow ink in mode AbsCol'. Settings for ColorGate Productionserver 4: RGB: sRGB for raster and vector, AbsCol CMYK: ISO Coated for raster and vector, AbsCol Gray: ISO Coated for raster and vector, AbsCol Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Tom Lianza Guest	Posted: Tue Jun 26, 2007 4:21 pm    Post subject: Re: Printing a Target by Absolute Colorimetry Quote: Now the questions: 1) How should a RIP treat Lab files ? Who knows a RIP with Lab input definition (besides spot colors) ? 2) How should the target be coded for correct printing ? Perhaps as an RGB file for an RGB space with D50 white point, and then printed by AbsCol ? Best regards --Gernot Hoffmann Hello Gernot, Strictly speaking, a Lab file measurement is "white" normalized. It's impossible to extract the assumed white due to the effective adaptation via Wrong Von Kries. This is the same normalization used in the early ICC profiles to form what was termed relative trichromatic coordinates. Absolute Color is really not meaningful for Lab scaled data. If you do not have apriori knowledge of the white point used for normalization, you can't get back to absolute XYZ. If the normalization used in your bluish target was a specified white such as D50, rather than paper "white", you should have no a*& b* values at zero. In that case, there should be color everywhere on the page. The page should be rendered with a D50 assumption because that is the normal interpretation of white in an Lab scaling. In this case, the paper white has some color when rendered into the absolute space. If the OUTPUT paper has a "white" point that matches this input, then there will be no ink on paper. If you are printing on a bright white medium, you should see some ink in the white if the output profile is properly formed. I think your suggestion #2 is quite rational and it should work. The colors in that file should be completely defined and unambiguous. That would be a good target to test the color management system AbsCol section of the profile. Regards, Tom
Back to top
	Ads
Advertising Sponsor
Gernot Hoffmann Guest	Posted: Wed Jun 27, 2007 11:02 am    Post subject: Re: Printing a Target by Absolute Colorimetry Tom Lianza schrieb: Quote: Hello Gernot, Strictly speaking, a Lab file measurement is "white" normalized. It's impossible to extract the assumed white due to the effective adaptation via Wrong Von Kries. This is the same normalization used in the early ICC profiles to form what was termed relative trichromatic coordinates. Absolute Color is really not meaningful for Lab scaled data. If you do not have apriori knowledge of the white point used for normalization, you can't get back to absolute XYZ. If the normalization used in your bluish target was a specified white such as D50, rather than paper "white", you should have no a*& b* values at zero. In that case, there should be color everywhere on the page. The page should be rendered with a D50 assumption because that is the normal interpretation of white in an Lab scaling. In this case, the paper white has some color when rendered into the absolute space. If the OUTPUT paper has a "white" point that matches this input, then there will be no ink on paper. If you are printing on a bright white medium, you should see some ink in the white if the output profile is properly formed. I think your suggestion #2 is quite rational and it should work. The colors in that file should be completely defined and unambiguous. That would be a good target to test the color management system AbsCol section of the profile. Regards, Tom Tom, thanks for the comments. In the moment I would like to discuss only one task: how to print a Lab-encoded target by an inkjet, using the RIP Colorgate Productionserver. 1. Lab encoding was done in an EPS program. In PostScript (PS) I'm defining the reference white as D50. PS is not the least ambiguous [funny enough, in the PostScript Language Reference, Third Edition, we can find a tutorial example with reference white D65]. PS Lab refers strictly to CIE XYZ (the PS connection space). The EPS was placed in a doc and exported as PDF. This is correct, as tested by Photoshop. 2. I had written in my first post: Printing PDF-Lab Standard settings sRGB / ISO Coated / ISO Coated, each RelCol, deliver visually correct color patches, but the gray patches are greenish. Settings Default / Default / Default, each AbsCol, deliver a pretty good appearance with neutral grays, but the measured values are slightly wrong (worse than for PDF-sRGB), and white paper is not covered by yellow ink. The RIP doesn't have an input Lab. It's rather strange, that RGB, CMYK and/or Gray profiles have an influence on the printed result for Lab input. My conclusion is now, that this isn't a question for this forum but for Colorgate. We know how the Lab AbsCol print should look: blue-ish paper is covered by yellow ink, and the measured Lab values of the patches should be near to the target numbers, as measured by DTP22 in mode 'absolute'. Best regards --Gernot Hoffmann
Back to top
	Ads
Advertising Sponsor
Display posts from previous: All Posts1 Day7 Days2 Weeks1 Month3 Months6 Months1 Year Oldest FirstNewest First

	Physics Talk Forum Index -> Color and Light	All times are GMT
Page 1 of 1

Jump to: Select a forum Physics Talk----------------Physics Talk-- BiophysicsMechanicsColor and LightRelativity  You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum

97 Attacks blocked

Powered by phpBB © 2001, 2005 phpBB Group