Getting Blue Colors "right"

[kenoli]

Most of my printing involves the printing of artwork, both reproductions of artwork and working with an artist to modify a hand painted original.

I have a persistent problem matching blue colors, especially a large area of blue, like a sky.

There seems to be a shift from "turquoisish" to "purplish", i.e. blue + green to blue + red.

I am generally working with a scanned image and am printing to an Epson P9000, though I have had exactly the same experience with an Epson 9900.

I typically make printer and scanner profiles using a X-rite i1Pro spectrometer and software.

I am not completely sure if the problem arises in the scanning or the printing, but I think it is in the printing.  I have my monitor profiled pretty well and generally the printed output corresponds fairly well with the color of the image on the monitor.

I can usually correct this (though it takes some work) using a combination of hue/saturation and levels in photoshop.  Sometimes it requires adjusting the blues in hue saturation and sometimes the cyan.  In levels, it involves adjusting the blue, of course.

It is always a challenging and persistent issue and I haven't found a set of layer settings that I can apply universally.   

I'm wondering if others have any relevant experience with this issue that they can share with me.

One friend who has this trouble thinks it is due to the lack of any "true" blue ink in the Epson ink lineup.  Epson does include orange and green inks and with the P9000 commercial version includes a violet that they say expands the blue gamut some, though we declined the violet ink as it requires switching out the LLK ink permanently which Epson says is not a good idea.  The Violet was added for proofing applications where the pantone colors have to be matched exactly.

I am wondering what result I could have gotten if I did use the violet option.

I have a pretty good set of scanner and printer profiles that reproduce most other colors fairly accurately, with little required color correction.

Any help would be appreciated.

--Kenoli

[Mark D Segal]

These Epson printers can produce any hue and tone of blue that you require. That isn't the problem. It sounds like a profiling or a scanning issue. What was the original media? What scanner and scanning software are you using? How was the scanner profiled in respect to the media you are using? Is the matching problem coming from the image you see on the monitor in your printing application versus the image from the printer, or is it what you see on the original media versus what you see from the printer?

[Doug Gray]

Most of my printing involves the printing of artwork, both reproductions of artwork and working with an artist to modify a hand painted original.

I have a persistent problem matching blue colors, especially a large area of blue, like a sky.

There seems to be a shift from "turquoisish" to "purplish", i.e. blue + green to blue + red.

I am generally working with a scanned image and am printing to an Epson P9000, though I have had exactly the same experience with an Epson 9900.

I typically make printer and scanner profiles using a X-rite i1Pro spectrometer and software.

I am not completely sure if the problem arises in the scanning or the printing, but I think it is in the printing.  I have my monitor profiled pretty well and generally the printed output corresponds fairly well with the color of the image on the monitor.

I can usually correct this (though it takes some work) using a combination of hue/saturation and levels in photoshop.  Sometimes it requires adjusting the blues in hue saturation and sometimes the cyan.  In levels, it involves adjusting the blue, of course.

It is always a challenging and persistent issue and I haven't found a set of layer settings that I can apply universally.   

I'm wondering if others have any relevant experience with this issue that they can share with me.

One friend who has this trouble thinks it is due to the lack of any "true" blue ink in the Epson ink lineup.  Epson does include orange and green inks and with the P9000 commercial version includes a violet that they say expands the blue gamut some, though we declined the violet ink as it requires switching out the LLK ink permanently which Epson says is not a good idea.  The Violet was added for proofing applications where the pantone colors have to be matched exactly.

I am wondering what result I could have gotten if I did use the violet option.

I have a pretty good set of scanner and printer profiles that reproduce most other colors fairly accurately, with little required color correction.

Any help would be appreciated.

--Kenoli

I agree with Mark. The printer is typically capable of printing colors accurately. The area where it is the most difficult to accurately render color is on the scanner side and it is easy to make or apply scanner profiles incorrectly.

A quick check is to scan a ColorChecker card, apply the profile, and load it in Photoshop. Then set the white point and luminance to match the lightest gray patch when read with the I1 Pro. The other patches should have Lab values that are at least reasonably close to that of the ColorChecker or a loaded reference image.

A simple and fast round trip test is to print an image of the ColorChecker from Bablecolor, scan it, then see how close the scanned image matches by looking at the actual Lab values in Photoshop.

I'm assuming you are using Absolute Colorimetry for reproduction work since this prints the actual colors while Relative Intent scales things so Lab(100,100,100) will be the actual white point of the media and that can be quite different. A Lab value of (50,0,0) should print and measure within +/-2 for each of the 3 numbers. Usually it is within +/-1.

[digitaldog]

These Epson printers can produce any hue and tone of blue that you require.

Blues shifting magenta is an old and rather common problem, just one wart found in Lab which is often used as the PCS. Lab assumes that hue and chroma can be treated separately yet, experimental results indicate that our perception of hue varies with color purity. Mixing white light with a monochromatic light does not produce a constant hue, but Lab assumes it does. This is noticable in how Lab deals with blues, and is the source of this blue-purple shift.

[Doug Gray]

Blues shifting magenta is an old and rather common problem, just one wart found in Lab which is often used as the PCS. Lab assumes that hue and chroma can be treated separately yet, experimental results indicate that our perception of hue varies with color purity. Mixing white light with a monochromatic light does not produce a constant hue, but Lab assumes it does. This is noticable in how Lab deals with blues, and is the source of this blue-purple shift.

While perceived hue changes with the luminance in the blues and red-oranges, the OP should not see such a change since the reproduction should match the artwork luminance it is a reproduction of.  That is, the reproduction and art Lab values should measure quite close. If they appear differently, then they will not measure very closely.

[Mark D Segal]

Andrew, yes I too have heard of that issue, but if this is a generic problem, why do I get perfectly satisfactory blue skies in the photos I make and print, whether scans from old film or from numerous digital cameras? And unless we have the OP's image to analyze (I don't see it anywhere) I'm not sure one can presume there is a generic problem at play here.

[Doug Gray]

While perceived hue changes with the luminance saturation in the blues and red-oranges, the OP should not see such a change since the reproduction should match the artwork luminance saturation it is a reproduction of.  That is, the reproduction and art Lab values should measure quite close. If they appear differently, then they will not measure very closely.

EtoA:  Corrected. The hue shift perceptual effect is relative to color saturation.

[digitaldog]

Andrew, yes I too have heard of that issue, but if this is a generic problem, why do I get perfectly satisfactory blue skies in the photos I make and print, whether scans from old film or from numerous digital cameras?

Because we have better color engines that attempt with some success to account for this old, blue-magenta shifting.

[Mark D Segal]

OK, so if the OP also benefits from the use of these better colour engines, it tells me we are back to a specific issue he is having, but until we know some answers to the questions I put, it's hard to know how to isolate his specific problem; anyhow, Doug Grey's analytical suggestions would make useful starting points.

[digitaldog]

Yes, we need a very low rez version of the image in it's working space to load into ColorThink, build a ColorList, plot it's gamut and run it through the output profile too. This isn't much different from the other thread going on here with issues reproducing blues. In that example, we (I) got an image in the working space and was able to plot it and show it's way outside the gamut of the Epson printer.

[Mark D Segal]

Yes, we need a very low rez version of the image in it's working space to load into ColorThink, build a ColorList, plot it's gamut and run it through the output profile too. This isn't much different from the other thread going on here with issues reproducing blues. In that example, we (I) got an image in the working space and was able to plot it and show it's way outside the gamut of the Epson printer.

No doubt you were able to make that determination, but at the same time, all these pro printers have enough gamut to reproduce very satisfactory blue skies; could it be that he needs to play with the R.I? Perhaps between RC and Perceptual there would be a substantial difference in the handling of the OOG blues?

[digitaldog]

No doubt you were able to make that determination, but at the same time, all these pro printers have enough gamut to reproduce very satisfactory blue skies; could it be that he needs to play with the R.I? Perhaps between RC and Perceptual there would be a substantial difference in the handling of the OOG blues?

That's certainly a possibility! Certainly with a perceptual rendering intent.

[GWGill]

Blues shifting magenta is an old and rather common problem, just one wart found in Lab which is often used as the PCS.

Only if you are clipping at L*a*b* constant hue lines. If this is within gamut, then L*a*b* is not the explanation.

[Doug Gray]

I think the first order of business is to clarify whether the OP is printing scene referred images. This is done for accurate reproduction work and is the norm for scanning then printing, but it is pretty awful for normal photography.  The  blue hue shift is a real issue for the latter because normally colors are more saturated with more contrast and that can cause a hue shift.

However, none of this applies to reproduction work. A Lab(30,0,-30) will be perceived to have a hue shift compared to a Lab(30,0,-60) whether they are printed or on the original artwork. The Lab values on reproduced art should be made as close as possible to the original art. The only time a hue shift difference between the two is when a color is outside the printable gamut and is mapped to something printable. That could be happening but my guess is that the major issue is the scanner profiling and application of the profile. This is an area where it is easy to get it wrong. It isn't as simple or accurate as making and using printer profile.

[digitaldog]

I think the first order of business is to clarify whether the OP is printing scene referred images. This is done for accurate reproduction work and is the norm for scanning then printing, but it is pretty awful for normal photography.

http://www.lumita.com/site_media/work/whitepapers/files/pscs3_rendering_image.pdf
By Karl Lang


Page 3:
The capture of scene-referred data is a science. The printing of output-referred data is also a science. The translation of an image from scene-referred to output-referred is the art of photographic printing. The original scene must be interpreted to produce a print. This process is subjective; there is no single “right way” to do this.


Page 7:
If you send scene-referred data directly to a printer, or even a high quality computer display, you will not be very happy with the result. Even using the lightness-mapped data will provide a at, lifeless image. These media don’t have the dynamic range that’s encoded in the original scene; in fact, far from it.


Page 8
Trying to reproduce the dynamic range of our original scene-referred image within the dynamic range of print isn’t possible.
In order to create a photographic image that evokes some resemblance to the original scene, we need to compress some parts of the tone scale and stretch others. We may also use clipping on at least one end. Clearly this process is both subjective and image-dependent. There is no universal right way, no formula or curve that will work for every image.

[GWGill]

I have a persistent problem matching blue colors, especially a large area of blue, like a sky.

There seems to be a shift from "turquoisish" to "purplish", i.e. blue + green to blue + red.

This sort of problem is usually pretty straightforward to figure out, if a little tedious. Track the color through your workflow.
[ Your biggest problem may be that you don't actually know what your workflow is in a technical, color numbers sense, and/or you are unfamiliar with the tools needed to do this. ]

i.e. Pick a uniform reference patch in the original. Measure it with your instrument. Scan it with your scanner.
Check the delta E and the error direction between what you measured and what the scanner profile predicts it will be using Absolute colorimetric. If that looks OK, move to the next step in you workflow.

If you are doing (say) a relative colorimetric to relative colorimetric conversion from scanner to printer then:

Convert the scanner RGB to relative (PCS) L*a*b* using the scanner profile relative colorimetric forward table. Convert that number to the printer "RGB" space using the printer profile relative colorimetric backwards table.

Check the printer profile invertibility/clipping by running that RGB forward through the printer profile using relative colorimetric, and check the resulting L*a*b* against the one you put in above.

Print a patch with that RGB. Measure it with your instrument.

Use the printer profile to convert the printer RGB to Absolute colorimetric L*a*b* using the forward table, and check the delta E to what you measured on the printer.

That's about it really. The source of your problems should show up there somewhere.
If the above indicates there is no problem (the delta E's are small and don't compound), then perhaps there is a difference between your actual workflow and the above manual one. You can do a similar exercise by comparing various values between files through your actual workflow and the manually calculated numbers.

Typically I would use ArgyllCMS icclu to do these profile conversions (of course!) - but maybe other tools could be used if they are more familiar (ColorThink ? Photoshop if you configure it properly ?).

[Mark D Segal]

Going back to square one for a moment, the OP still hasn't told us what the original media is, what scanner is being used and where the profiles come from. The original media is important to know about because there is a substantial difference in the handling especially of Blue between Kodachrome and all other chromes. If scanning Kodachrome, it is best to use specific profiles for Kodachrome. To makes those, a Kodachrome target is needed. Those are now as scare as hen's teeth and cost a fortune. I believe there are some Kodachrome emulation algorithms available that circumvent this problem. Between flavours of other chromes there are differences of colour rendition that are best served with bespoke targets. Then there is the scanner. Depending on the quality of the scanner, the consistency and accuracy of colour rendition could be an issue as well. A number of variables could be affecting the colour matching between the painting and the print right from the scan stage, before one even gets to how OOG colours are handled. It would be helpful when those looking for assistance provide as much relevant information as possible. It makes proposing useful answers more prospective.

[Doug Gray]

http://www.lumita.com/site_media/work/whitepapers/files/pscs3_rendering_image.pdf
By Karl Lang


Page 3:
The capture of scene-referred data is a science. The printing of output-referred data is also a science. The translation of an image from scene-referred to output-referred is the art of photographic printing. The original scene must be interpreted to produce a print. This process is subjective; there is no single “right way” to do this.


Page 7:
If you send scene-referred data directly to a printer, or even a high quality computer display, you will not be very happy with the result. Even using the lightness-mapped data will provide a at, lifeless image. These media don’t have the dynamic range that’s encoded in the original scene; in fact, far from it.


Page 8
Trying to reproduce the dynamic range of our original scene-referred image within the dynamic range of print isn’t possible.
In order to create a photographic image that evokes some resemblance to the original scene, we need to compress some parts of the tone scale and stretch others. We may also use clipping on at least one end. Clearly this process is both subjective and image-dependent. There is no universal right way, no formula or curve that will work for every image.

I agree with all of that so your point is what?

You do realize (I hope) that the process of scanning and printing to reproduce the scanned article is done using a scene referred process.

If a friend gives you a 4x5 borderless print and asks you to enlarge it to an 8x10 the proper way to do it is to scan it with a scene referred profile. Resize it. Then print it using Absolute Colorimetric Intent.  As long as the print media has a gamut that is as large as that in the scanned image there is no art to it. Just the science and the numbers.

The OP is scanning something and wishes to reproduce it. A good scanner profile and AC printing is the best way to do it.

[Mark D Segal]

I agree with all of that so your point is what?

You do realize (I hope) that the process of scanning and printing to reproduce the scanned article is done using a scene referred process.

If a friend gives you a 4x5 borderless print and asks you to enlarge it to an 8x10 the proper way to do it is to scan it with a scene referred profile. Resize it. Then print it using Absolute Colorimetric Intent.  As long as the print media has a gamut that is as large as that in the scanned image there is no art to it. Just the science and the numbers.

The OP is scanning something and wishes to reproduce it. A good scanner profile and AC printing is the best way to do it.

Yes, and I think this makes all the more sense insofar as the OP indicates these are scans and prints of photographed artwork - (likely paintings?), which would have far fewer gamut issues than photographs of the great outdoors. At least blue values in the original art should be generally amenable to measurement, which would indicate the existence of potential gamut compression issues straight away, once compared with the gamut limitations of the printer/paper profile discernible in Color Think Pro.

[Doug Gray]

Going back to square one for a moment, the OP still hasn't told us what the original media is, what scanner is being used and where the profiles come from. The original media is important to know about because there is a substantial difference in the handling especially of Blue between Kodachrome and all other chromes. If scanning Kodachrome, it is best to use specific profiles for Kodachrome. To makes those, a Kodachrome target is needed. Those are now as scare as hen's teeth and cost a fortune. I believe there are some Kodachrome emulation algorithms available that circumvent this problem. Between flavours of other chromes there are differences of colour rendition that are best served with bespoke targets. Then there is the scanner. Depending on the quality of the scanner, the consistency and accuracy of colour rendition could be an issue as well. A number of variables could be affecting the colour matching between the painting and the print right from the scan stage, before one even gets to how OOG colours are handled. It would be helpful when those looking for assistance provide as much relevant information as possible. It makes proposing useful answers more prospective.

This is very true for the most precise colorimetric replication.  Other than that you just have to spot check colors with a spectro and hope you're are close enough.