Problem: Copying Artwork & Matching Colours

[Stephen G]

Edit: ignore this post. pure nonsense. I found I could simulate the blue outcome in softproof, but there's no way to do this in a print dialogue, so my hunch that the OP was hitting an incorrect setting somewhere is a waste of time.

Looking at the two images in the original post I tried to simulate the blue outcome in PS by messing around with assigning profiles and trying various soft-proofing settings.

Questions for the OP:
Are you printing through PS?
If so, is the 'Preserve RGB numbers' box perhaps checked?

If not then I'm out of ideas for now.

[William Walker]

I doubt the problem is on the printer side. It sounds more like the photographed image is being processed using the normal Output Referenced process. For artwork reproduction you need a Scene Referenced (except for white point) DCP or ICC and the image has to start off as a RAW image. Camera images that are generated as jpegs are virtually always Output Referenced and have S-curves applied for dynamic range reduction. This will never produce good images for reproduction.

Hi Doug

I am not sure what "Output Referenced" and "Scene Referenced" processes are...to be clear, I am using a Canon 5dSR, RAW and ProPhoto Colour Space. The artwork was lit by two 5500k "Neutral daylight" softboxes at 45 degrees to the artwork and the colours were matched from the display to the artwork under the same lights.

Marcin, thanks for the link to the Doyle's patch - Before I do anything, does this screenshot tell you whether I need to use the patch?

Thank you all for your input so far!

[Doug Gray]

Hi Doug

I am not sure what "Output Referenced" and "Scene Referenced" processes are...to be clear, I am using a Canon 5dSR, RAW and ProPhoto Colour Space. The artwork was lit by two 5500k "Neutral daylight" softboxes at 45 degrees to the artwork and the colours were matched from the display to the artwork under the same lights.

Marcin, thanks for the link to the Doyle's patch - Before I do anything, does this screenshot tell you whether I need to use the patch?

Thank you all for your input so far!

When you Scene Refer and convert a RAW file you retain colorimetric values and are limited only by the gamut boundaries in whatever colorspace you choose. Conventionally you are allowed to white balance but that's all.

As an example if you had a camera with a perfect (theoretical only) CRI (color rendering index) you could take a picture of a ColorChecker (or anything else) in D50, Convert the RAW image as Scene Referenced to ProPhoto and look at the Lab values of each of the squares and they would match, within instrumentation error, the measured values of the patches.

If you took the same RAW image and converted it as Output Referenced, i.e., images not intended for repro purposes, then both colors and tone curve would be significantly different. When printed, the second whitest patch would be brighter due to "S curve" compression. Same in the opposite direction for the blackest patch.


http://www.color.org/scene-referred.xalter

A scene-referred image is an image where the image data is an encoding of the colors of a scene (relative to each other), as opposed to a picture of a scene. In a picture, the colors are typically altered to make them more pleasing to viewers when viewed using some target medium.

The emergence of a wider variety of reproduction media, and the penetration of digital workflows into virtually all aspects of imaging have led to increased interest in the exchange of camera raw and scene-referred images. Such images can subsequently be color rendered to different reproduction media, optimizing the image for each. Camera raw and scene-referred workflows are fundamentally different from output-referred workflows, such as those based on sRGB or Adobe RGB (1998) image exchange.

[digitaldog]

I am not sure what "Output Referenced" and "Scene Referenced" processes are...

http://www.color.org/ICC_white_paper_20_Digital_photography_color_management_basics.pdf

[Czornyj]

Marcin, thanks for the link to the Doyle's patch - Before I do anything, does this screenshot tell you whether I need to use the patch?

Everything seems to be ok, so apparently it's not a matter of this issue and you don't need a patch.

[William Walker]

Everything seems to be ok, so apparently it's not a matter of this issue and you don't need a patch.

Thanks! I followed the "Scene Referenced" instructions as per the ICC white paper - the colours appear (on the display) pretty close to the original painting - the problem seems to be in the printing. I will try a canned profile.

[BradSmith]

Something basic, rather than esoteric is wrong.  Those colors might be (probably are) out of gamut, but bringing them back into gamut would be much, much closer to the original compared with what your results are.   in this case, those colors in your two images are not off a little and are unacceptable.....they are a million miles apart.   
Brad

[Mark D Segal]

Something basic, rather than esoteric is wrong.  Those colors might be (probably are) out of gamut, but bringing them back into gamut would be much, much closer to the original compared with what your results are.   in this case, those colors in your two images are not off a little and are unacceptable.....they are a million miles apart.   
Brad

Yes, I agree, something basic; but what? It isn't clear yet whether the problem is at the capture end or the printing end of the workflow. It could be either, though there was some evidence it may be at the printing end (the screen match between image and painting). A friend is having a similar issue and in his case it turns out that a custom camera profile made with BasicColor Input and very careful attention to lighting has gone a very long way to solving his problem. Doesn't necessarily mean the same applies to this case. The OP needs to do more stage by stage analysis to ascertain the real cause; then a solution would be closer at hand.

[BobShaw]

I have just had a frustrating day trying to make a print from a painting.
2. I then made a few small adjustments, with the painting still illuminated, on my calibrated NEC PA272W. I made an accurate on-screen match with the painting.

What exactly do you mean by that? It sounds like you uncalibrated a previously calibrated monitor to make it match your uncalibrated camera.

The camera profile converts RGB to LAB numbers in the computer. The monitor profile then converts those LAB numbers in the computer to an RGB value on the screen.
If the LAB numbers captured were wrong then adjusting the monitor makes it worse.

[kirkt]

Take the display itself out of the equation for a moment and simply print a reference test print such as those, for example, on Keith Cooper's or Andrew Rodney's websites:

http://www.northlight-images.co.uk/article_pages/test_images.html

or

http://www.digitaldog.net/files/Gamut_Test_File_Flat.tif

for example.  These kinds of files are constructed to be "correct" and test the capabilities of your printer and print workflow - depending upon which you select, there may be colors in the reference file that are out of gamut for your printer.  However, you can print a reference print without even displaying it on your monitor, the point being that it should print "correctly" without your intervention, editing, worrying about color balance, etc.  If you print it and then compare it to your display, and one looks way out of whack compared to the other, then you can start to isolate the problem without speculating as to the effects of your workflow on the output of each device.

It may be "better" to select or contrive a test file that is in sRGB so that the colors in the image file will fit into the gamut of both your display and your printer (within reason), eliminating the issues of way out of gamut color from confounding your troubleshooting.

Also, while not exactly germane to your specific workflow, here is a worthwhile read regarding repro imaging and the under-the-hood mechanics of raw conversion for artwork.

http://www.trumpetpower.com/photos/Exposure

good luck,

kirk

[AlterEgo]

The camera profile converts RGB to LAB numbers in the computer.

it actually depends on raw converter... some don't use lab anywhere in color transform related processing all, but use for other purposes - like masking - ACR/LR for example

[William Walker]

What exactly do you mean by that? It sounds like you uncalibrated a previously calibrated monitor to make it match your uncalibrated camera.

The camera profile converts RGB to LAB numbers in the computer. The monitor profile then converts those LAB numbers in the computer to an RGB value on the screen.
If the LAB numbers captured were wrong then adjusting the monitor makes it worse.

No Bob, what I meant to say was that my monitor is properly calibrated, the tweaks I made were in Lightroom, so that the the colours on the (properly calibrated) display matched the colours of the painting.

The print is where I am battling to match the colours on both the display and the painting itself.

Keep in mind that it is only with this particular image that I am having a problem. My "normal' day-to-day printing of my landscape photography and for other people for who I print are fine.

I digitised 22 pieces of artwork for another artist and had a similar problem with 2 of those, the rest were fine. The problem seems to be in the green/blue areas. It just seems crazy that my prints normally get pretty close to what I see on the display, but not in this instance...

[kirkt]

...

I digitised 22 pieces of artwork for another artist and had a similar problem with 2 of those, the rest were fine. The problem seems to be in the green/blue areas. It just seems crazy that my prints normally get pretty close to what I see on the display, but not in this instance...

Do you mind providing link to download the custom profile you made for your printer+paper combination?

kirk

[Bart_van_der_Wolf]

The print is where I am battling to match the colours on both the display and the painting itself.

Keep in mind that it is only with this particular image that I am having a problem. My "normal' day-to-day printing of my landscape photography and for other people for who I print are fine.

I digitised 22 pieces of artwork for another artist and had a similar problem with 2 of those, the rest were fine. The problem seems to be in the green/blue areas. It just seems crazy that my prints normally get pretty close to what I see on the display, but not in this instance...

Hi,

I suggested the printing of a Granger rainbow test image, as it would indicate if there is a gap or other discontinuity in the blue-green colour rendering by the output profile. If there is, then you've found the issue (and need another profile for the specific medium). Do try printing with different Rendering intents, they will make different trade-offs.

Cheers,
Bart

[William Walker]

Everything seems to be ok, so apparently it's not a matter of this issue and you don't need a patch.

Hi Marcin - when I try to print from Colormunki to create a new profile, I get these "options"...(this is with Doyle's Patch) which are greyed-out.

Is this my problem?

Thanks
William

[Czornyj]

Hi Marcin - when I try to print from Colormunki to create a new profile, I get these "options"...(this is with Doyle's Patch) which are greyed-out.

Is this my problem?

Thanks
William

Will,

Color Mode: greyed-out "Color" indicates that the printed targets are color managed by printer driver so yes, there's something wrong, there should be "No Color Correction".

Try saving the targets and print them from PS using Photoshop iPF PrintPlugin:

[xpatUSA]

The jade in the reference sRGB image has a clipped red channel to zero, converting to AdobeRGB shows no clipped numbers.

The over-saturation (clipped reds) in both posted images seems to be similar, according to Show Image.

Here's the saturation planes compared:

http://kronometric.org/phot/post/LL/compSat.jpg

The hues are quite different though. Might be a clue there:

http://kronometric.org/phot/post/LL/compHue.jpg

Rgds,

[Tim Lookingbill]

The over-saturation (clipped reds) in both posted images seems to be similar, according to Show Image.

Here's the saturation planes compared:

http://kronometric.org/phot/post/LL/compSat.jpg

The hues are quite different though. Might be a clue there:

http://kronometric.org/phot/post/LL/compHue.jpg

Rgds,

Have you printed a crop of the jade color in the OP's posted jpeg of the source file on your printer using similar hot press type paper/profile, or just printing with "Printer Manages Color"? This is a tagged image that should communicate to the printer what ink amount combination gets dropped on the paper to reproduce that jade color. The numbers don't matter of course.

I only made that point to indicate that particular hue of jade may or may not be right on the edge of the printer's color gamut capabilities to where there may now be a very narrow space on the target to get it or not get it on paper. Can't recall in this thread if someone mapped that jade to see if it fits within that particular printer/paper combo reproduction capability.

[William Walker]

I would like to start off by thanking everyone who has helped me!

I have been quite busy this week and have not been able to respond as quickly as I would have liked.

I have spent the most of today trying to make some progress...

A.  I made new profiles for three different papers using Simon Simpson's PDF work-around (http://forum.luminous-landscape.com/index.php?topic=91177.0) and have not had any success. When soft-proofing, the picture becomes very "muddy" - awful. The print then ends up looking nothing like neither the original nor the soft-proof!
B.  I tried a glossy Canon photo paper with both my profile and the canned profile and ended up with the same blues that I started with originally.
C.  I printed the Granger rainbow as per Bart's instructions (Perceptual frist and then Relative) and have posted the results below. What do you make of that Bart? Clearly a problem there. (NB!! That is on the Canon Photo Paper with the canned profile.)
D.  Finally, as a long shot, I used some cheap coated paper with a Canon "Coated Paper" Canned Profile and got as close to the jade as to be just about perfect!???

I am not sure now if I am any closer to an answer...perhaps someone can pick up something from this?

Thanks again for your time! I really appreciate it!

William

[Bart_van_der_Wolf]

C.  I printed the Granger rainbow as per Bart's instructions (Perceptual frist and then Relative) and have posted the results below. What do you make of that Bart? Clearly a problem there. (NB!! That is on the Canon Photo Paper with the canned profile.)

Hi William,

As I somewhat expected, there is a profile issue in that Blue/Green/Jade region. The Relative Rendering intent is a bit smoother than Perceptual, but not real smooth yet. So I suspect a reprofiling might be at least a partial solution, since camera sensitivity curves do not play a role in the Granger Rainbow.

It's still a bit strange that the resulting color shift is so drastic, but it might be tied to the profile issue.

Cheers,
Bart