Pages: 1 [2]   Go Down

Author Topic: Getting Blue Colors "right"  (Read 9228 times)

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Getting Blue Colors "right"
« Reply #20 on: December 22, 2015, 08:15:49 pm »

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.

You got me thinking a bit. Because of the limited dye set for wet processes, the math to fix up the scanner output should be fairly simple and you don't need many patches because the spectral characteristics should be just a linear combination of the dyes. It's a much simpler problem than trying to fit a CFA response to a large number of natural colors. Simple regression should produce quite accurate results. Likely a simple matrix would map the colors correctly to D50. At least if there are only 3 dyes in the wet process. If there are more dyes than that one would probably need to use LUTs.

A simple picture of a colorchecker printed using the wet process would be more than enough. One measures the print with a spectro and that becomes the target to extract the dye responses from.
« Last Edit: December 22, 2015, 08:19:55 pm by Doug Gray »
Logged

kenoli

  • Newbie
  • *
  • Offline Offline
  • Posts: 45
Re: Getting Blue Colors "right"
« Reply #21 on: December 22, 2015, 09:42:07 pm »

Wow!  What an amazing array of responses.  It is going to take some sorting through to give them justice, and, I think a bit of learning on my part to fully understand all of them.

I am scanning with an Epson 10000XL.  I have made scanner profiles using the i1Pro device & software and with Vuescan and, in the end have found that I get the best results with the profile that Apple loads for that scanner. 

I am scanning reflective RGB and scanning a variety of sources, mostly original artwork and it is the latter that offers this challenge.  The artwork I am working with currently involves a few media, including traditional watercolor paintings on watercolor paper, mixed media on flat matte paper made up of a range of media from oil pastel to pan pastel to collages to watercolor.  The fields of blue watercolor in the mixed media and the traditional watercolor are the areas that give me the most trouble.

I will try some of the tests proposed, to the point I actually understand how to do them and report back.  I may not get to it until after Christmas.  I do have a color checker target and a collection of Wolf Faust targets that I use for making scanner profiles.  Vuescan actually offers a process for making profiles all the way from the scanner to the printer that I haven't tried.

I greatly appreciate the thought and time all of you have put into this.

--Kenoli (the OP)
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Getting Blue Colors "right"
« Reply #22 on: December 22, 2015, 10:07:36 pm »

OK, this is starting to clarify. You can have difficulty with this kind of scanning because the reflective properties of paint on paper seem to differ enough from what the profiling targets' colors have been configured for that accurate profiling becomes very difficult. I have heard of this problem before and the only viable solution I've read has been workarounds using editing tools either in the scanner software or in the post-scan editing process. (I know one technically savvy photographer/artist who tried to produce a profiling target using the pigments of the painted media, but as I recollect it was not a good solution - perhaps I can check with him and refresh my memory of this - it was years ago.) Hence the absence of a good profiling solution for this particular kind of scanning adds manual intervention and time, but off hand I don't know of a viable alternative.

I assume of course that you were using a reflective target to make a profile for a reflective scanning process. That is essential, if it is to work at all.

You may not be satisfied with the quality of results from a Pro-level printer if you are using a scanner as the reader of the printer profiling target, if that is what you meant by Vuescan offering a complete profiling solution from scan to print. So if you haven't spent time on this, that's fine.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Getting Blue Colors "right"
« Reply #23 on: December 23, 2015, 01:20:07 am »

I'm sure he isn't using the scanner to profile his printer. He said he uses an I1Pro and software for making scanner and printer profiles.

However, applying scanner profiles correctly is tricky to the point I gave up and wound up just assigning the scanner profile after loading the unmanaged scan in Photoshop. One way too make sure the profile is used properly.

Your points about the difficulty scanning media with irregular surfaces, especially lumps with glossy surfaces. Profiles are all made with 0/45 spectros and I have no idea how well the scanner replicates that geometry. And the artwork surface geometry is another factor to make life difficult. Watercolor art should be ok though.  Probably his best bet is to check the color accuracy by following GWGill's post. Once that is verified he should be able to see whether the scanner will be adequate for artwork.
Logged

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Getting Blue Colors "right"
« Reply #24 on: December 23, 2015, 07:12:34 am »

I'm sure he isn't using the scanner to profile his printer. He said he uses an I1Pro and software for making scanner and printer profiles.

However, applying scanner profiles correctly is tricky to the point I gave up and wound up just assigning the scanner profile after loading the unmanaged scan in Photoshop. One way too make sure the profile is used properly.

Your points about the difficulty scanning media with irregular surfaces, especially lumps with glossy surfaces. Profiles are all made with 0/45 spectros and I have no idea how well the scanner replicates that geometry. And the artwork surface geometry is another factor to make life difficult. Watercolor art should be ok though.  Probably his best bet is to check the color accuracy by following GWGill's post. Once that is verified he should be able to see whether the scanner will be adequate for artwork.

True Doug, he is not using the scanner to profile his printer. He just mentioned what appears to be that that as an option in Vuescan and I responded that it's iffy in principle.

I think the character of the reflectance (influenced by the factors you mention here) plays a role in how well a scanner and profile combination can deal with colours in painted media. I agree one should expect watercolour to be less challenging than some other paint media, but not necessarily - could depend on the kind of watercolour and what it is painted on.

The workflow options of scanning with the correct profile at the scan stage, or applying the same scanner profile to a "raw" scan in Photoshop produces identical results from my testing, provided the scanning application itself is amenable to good colour management. As you would imagine, I use SilverFast Ai Studio 8.x and have found either option equivalent.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Mark D Segal

  • Contributor
  • Sr. Member
  • *
  • Offline Offline
  • Posts: 12512
    • http://www.markdsegal.com
Re: Getting Blue Colors "right"
« Reply #25 on: December 23, 2015, 07:46:22 am »

Further on this subject, to refresh my memory of discussions with highly experienced and knowledgeable practitioners from years gone by, I pulled up this response from one of them who will remain anonymous "to protect the guilty"; we were talking about how to work around the problems of accurate painting reproduction using cameras or scanners with conventional profiles. I think these comments are relevant and insightful in this context:

<<Just a remark here about what we discussed a few years ago on the subject of  ** Using the actual artist's colors and building a profile based on those colors ** I'm talking specifically about watercolor ...
 
<<1- Measuring off the artist painting is pointless because the spectrophotometer scans only a minute part of the pigments that make up the color,  these are not representative of the color that our eyes perceive  (I did extensive tests of that and sent you samples of my tests, I'm not sure if I have kept them).
 
<<2- So you must photograph or scan the art then [ >Select ] a larger area of the color and  [Filter>Blur>Average] that color ... and then you are no longer measuring the *original*.  You are measuring a bad photograph displayed on a screen possibly using a bad profile ...
 
 
<<I *did* build sets of hundreds of patches (of diminishing intensity) using a brush and the *actual pigments from the artist*.  That did not help much because the artist in question always used a *mixture* of  two or more pigments together.
 
<<I am now of the opinion that if the camera could capture all the colors accurately and "equally" then an accurate reproduction would be achievable using Lightroom and Photoshop.
 
<<I agree that accurate lighting and good profiles are important but they will not get you close enough to an accurate reproduction of the painting unless you do a *lot* of tweaking with Photoshop.>>

The discussion back then focused on both scanning and camera capture, but more the latter; the basic problem posed by the reflectance characteristics of the original media was seen to be of a similar character.
Logged
Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....."

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Getting Blue Colors "right"
« Reply #26 on: December 23, 2015, 01:22:46 pm »

Further on this subject, to refresh my memory of discussions with highly experienced and knowledgeable practitioners from years gone by, I pulled up this response from one of them who will remain anonymous "to protect the guilty"; we were talking about how to work around the problems of accurate painting reproduction using cameras or scanners with conventional profiles. I think these comments are relevant and insightful in this context:

<<Just a remark here about what we discussed a few years ago on the subject of  ** Using the actual artist's colors and building a profile based on those colors ** I'm talking specifically about watercolor ...
 
<<1- Measuring off the artist painting is pointless because the spectrophotometer scans only a minute part of the pigments that make up the color,  these are not representative of the color that our eyes perceive  (I did extensive tests of that and sent you samples of my tests, I'm not sure if I have kept them).
 
<<2- So you must photograph or scan the art then [ >Select ] a larger area of the color and  [Filter>Blur>Average] that color ... and then you are no longer measuring the *original*.  You are measuring a bad photograph displayed on a screen possibly using a bad profile ...
 
 
<<I *did* build sets of hundreds of patches (of diminishing intensity) using a brush and the *actual pigments from the artist*.  That did not help much because the artist in question always used a *mixture* of  two or more pigments together.
 
<<I am now of the opinion that if the camera could capture all the colors accurately and "equally" then an accurate reproduction would be achievable using Lightroom and Photoshop.
 
<<I agree that accurate lighting and good profiles are important but they will not get you close enough to an accurate reproduction of the painting unless you do a *lot* of tweaking with Photoshop.>>

The discussion back then focused on both scanning and camera capture, but more the latter; the basic problem posed by the reflectance characteristics of the original media was seen to be of a similar character.

Most interesting.

I prefer to do reproductions using cameras rather than scanners for many of the reasons above.  The exception being where I need precise dimensional control. Photographs have to be dimensionally mapped and corrected for lens and illuminant levels. This is pretty easy for a fixed setup where you have a reference illuminance image. Then a short Matlab script does the adjustments.

These aside, what photography allows that scanners don't is control over the angles of illumination. My preferred setup is two lights on both sides at 45 degrees to the object as this closely matches the way profiles require spectrophotometer measures. This usually works but sometimes you can get reflections from surface features you won't see if, say, the art is displayed from above with a 30 degree illuminant angle. With cameras, you can change the illumination positions to avoid much of that.

The Luther condition is a problem with both scanners and cameras but more so with scanners because the combination of the color filters and the scanner's ragged spectrum makes color rendition worse.

I recall a paper that showed camera CFAs can be made to match Luther conditions much better by using 3 photographs, two with different colored filters. The RAW file color channels from each image can then be processed to provide much more accurate XYZ values.  It should work pretty well with the right filters but I never tried it and never really had the need for the increased accuracy the technique would provide.
Logged

digitaldog

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 20630
  • Andrew Rodney
    • http://www.digitaldog.net/
Re: Getting Blue Colors "right"
« Reply #27 on: December 23, 2015, 01:45:46 pm »

I know of no shipping camera that meets the Luther-Ives condition. This means that cameras exhibit significant observer metamerism compared to humans.
http://dougkerr.net/Pumpkin/articles/Metameric_Error.pdf
Logged
http://www.digitaldog.net/
Author "Color Management for Photographers".

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Getting Blue Colors "right"
« Reply #28 on: December 23, 2015, 02:32:42 pm »

I know of no shipping camera that meets the Luther-Ives condition. This means that cameras exhibit significant observer metamerism compared to humans.
http://dougkerr.net/Pumpkin/articles/Metameric_Error.pdf

That's right and it hasn't really improved over the years. The paper I was referring to used several color filters in addition to an unfiltered image to effectively improve closeness to Luther-Ives. It was published quite some time ago. I believe the person was working with the Rochester folks. It's been a while so I could be wrong on that.

I just read the article you linked. It's quite a good explanation and overview of the issue. Thanks.
« Last Edit: December 23, 2015, 02:40:56 pm by Doug Gray »
Logged

kenoli

  • Newbie
  • *
  • Offline Offline
  • Posts: 45
Re: Getting Blue Colors "right"
« Reply #29 on: January 06, 2016, 05:17:06 pm »

Everyone:

I continue to be overwhelmed by the thought and skill shown in the replies to my query.

I am greatly appreciative though not sure what next steps to take since a lot of this is beyond my knowledge level.  I'm fascinated by the conversation and would like to understand everything better than I do.

I would greatly appreciate a suggested process in words a much less knowledgable person (that is, me) can understand that would either help solve the problem or give you more information that would help move the discussion along towards a solution.  I'm willing and anxious to learn.  I do have a color checker target and did read through the babble color website, to some degree.  I'm not sure if I need their $250 software or not to follow suggestions related to their site.

I will add a bit of info here that might be helpful:

1.  I am perceiving the shift from blue (on the turquoise side) to blue (on the violet side) largely from what I see on the monitor to what I see coming out of the printer.  This occurs both from images I have scanned and images artists bring to me.  The primary artist involved, however, is bringing me images that he digitized with a scanner.  I do have his originals for comparison.  For others, I don't know the source of the images.

2.  I have carefully profiled the scanner, monitor and printer, all three with i1pro and the scanner with other devices as well.  The problem with the i1pro scanner profiling process is that it requires a scan of a target (I am using wolf targets) and I am not sure exactly how to get a scanned image that is not already affected by the scanner driver.  VueScan has a process they say will do this as well as the Epson Scan driver, though I get quite different results depending on which I use to provide the image to i1pro.  In the end, I have found that I get the best results with the scanner profile provided through Apple's operating system (not sure who provided it to them or how it was created).

3.  Other than the blue issue, I am getting pretty good color pass through all the way from the scanner to the printer.

4.  I have a t3i camera, and some soft boxes that provide lots of light.  I have decided to use the scanner instead of the camera as I commonly need much better resolution than I can get from the camera.  In the artwork I do with my partner, we commonly enlarge scans many times and images shot with the camera are inadequate.  I also had serious color challenges using the camera input, but that was before I really knew how to manage color inputs from a camera.  I now have some tools to help with that, including color checker and some Expodisc filters, though color checker is pretty much keyed to Lightroom, with which I have little experience.  The scanner has been a simple solution and I can scan large artwork in sections and stitch it together in Photoshop and have them at high resolution.

5.  I do have the Adobe Lightroom software (CS6), though I have only looked at it briefly and do not use it.  I have been using Aperture to organize my images but use Photoshop CS6 to edit images.  Everything is upside down in the Aperture world at this juncture.

I can provide any other info anyone would like.

Again, I appreciate the thought and time people have put into this.  I seem to have struck an interesting topic.

--Kenoli

Logged

Steve Upton

  • Newbie
  • *
  • Offline Offline
  • Posts: 23
Re: Getting Blue Colors "right"
« Reply #30 on: January 06, 2016, 09:03:21 pm »

I continue to be overwhelmed by the thought and skill shown in the replies to my query.

There are a few things you could try before it gets too complicated.

1. How's your lighting? If it's not controlled and it's fairly yellow (most indoor lighting is) then your blues may appear purple.

2. Work on isolating the problem. Try printing known-good images with similar blues and see if they shift. (Andrew Rodney has a great test image for such colors). Also try scanning a known good image (more difficult to get your hands on) to see if blues shift on it. Isolating the problem is the key.

Logged

Doug Gray

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2197
Re: Getting Blue Colors "right"
« Reply #31 on: January 06, 2016, 09:52:10 pm »

You have an I1 Pro.  First establish that your monitor and printer profiles are operating correctly. The monitor profiling software should ensure that.  Second, make a few blue/purple patches in an image in Photoshop using Lab color space. Print them out using Absolute Intent, then measure them with the I1 Pro.  The Lab values should be very close to the colors you made the patches in the image. The important thing is that the measurement of the patches match the Lab values you entered to create the color patches because their appearance can differ a lot due to illuminant differences from the monitor's setup.

Basically, you want to check each component of your workflow for accuracy. Andrew Rodney has some good tutorial videos on how to adjust monitors to match your print viewing. It's important to also do this if you want what you see on a monitor to match what you see on a print but the most important thing is to be able to print a defined color accurately and that has nothing to do with the monitor side. It's entirely a function of the printer and profile being set up correctly.

Once these are working then the next step should be to verify or fix the scanner profile.  Divide up the issues and address one at a time. Trying to get everything "right" without using instruments to measure each aspect separately, is a hard and frustrating process. 
Logged
Pages: 1 [2]   Go Up