Most Accurate Method to Manage Out of Gamut Color in Post

[Brad P]

For context, I often spend a day on individual images and print them myself at large sizes.  Like all of us, I try to reproduce color as accurately as possible. 

I am looking for a better practical method than simply switching on soft proofing when editing camera-profiled ProPhoto files in an otherwise properly color calibrated workflow.  Too often managing prints while in post to the rather blunt soft proofing out of gamut warnings results in desaturation, muted color detail and/or color shifts. Even using Photoshop's full array of color management tools, the feedback I get from the out of gamut warnings seems too crude.  And I don't want to waste paper with the trial by error print and adjust method. 

The best method I have found to further improve my post processing workflow may involve purchasing Chromix's ColorThink, analyzing individual files against my printer gamut as I go, and attempting to make more targeted adjustments.  Before I go to that route, is that about right or is there anything better?

[Mark D Segal]

I do 95% of my post capture editing in Lightroom under softproof, and use the histogram as guidance for when there is saturation or luminance clipping or both. I use the word "guidance" advisedly, because if one's colour management pipeline is correct and using a photo-quality display, such as an NEC or an Eizo, the soft-proof is usually the best arbiter of what to expect  out of the printer. I sometimes use the Color Worksheet function of ColorThink Pro to visualize which colours are out of gamut, but that doesn't say what one should do about it if there are any. By moving ones eyes back and forth between the photo and the histogram as one edits, the outcome will be as predictable and controllable as feasible with current technology.

[Doug Gray]

For context, I often spend a day on individual images and print them myself at large sizes.  Like all of us, I try to reproduce color as accurately as possible. 

I am looking for a better practical method than simply switching on soft proofing when editing camera-profiled ProPhoto files in an otherwise properly color calibrated workflow.  Too often managing prints while in post to the rather blunt soft proofing out of gamut warnings results in desaturation, muted color detail and/or color shifts. Even using Photoshop's full array of color management tools, the feedback I get from the out of gamut warnings seems too crude.  And I don't want to waste paper with the trial by error print and adjust method. 

The best method I have found to further improve my post processing workflow may involve purchasing Chromix's ColorThink, analyzing individual files against my printer gamut as I go, and attempting to make more targeted adjustments.  Before I go to that route, is that about right or is there anything better?

Couple of questions and observations. First, do you have a wide gamut monitor? Soft proofing is considerably better with one. It's useful with sRGB type monitors because there are significant numbers of colors in the monitor's sRGB gamut that can't be printed and those will show up with soft proofing. But wide gamut monitors are much better and cover the large bulk of printable colors. But not all. Particularly in the cyans.

Also, between PI and RI, only RI is colorimetric where a distinction between in gamut and out of gamut colors can be reasonably determined. PI smooshes colors near and somewhat beyond the printer's gamut so there is no distinct boundary. RI provides clear boundaries.

As for Photoshop's OOG masking. It's requires about 6 dE before it shows the mask. I have posted details of an investigation that shows that in another thread on LuLa. But under that and you won't see the mask.

The other issue is those colors that are outside the monitor's gamut. These are shown incorrectly in Photoshop because the color conversions typically increase the luminance while clipping at the gamut boundary while the printer profile, which has a different path, will not change the luminance until the printer's gamut boundary is reached. Further, printer profiles, unlike monitor profiles, typically do not significantly change the luminance beyond the gamut boundary.

One way to improve soft proofing in these situations is to desaturate the monitor's colors by 20% or so. This is a setting in Photoshop's edit->settings menu. This prevents clipping described above and improves softproofing at the cost of desaturating the entire image slightly. But it makes it easier to see problematic areas if your proof colors exceed what the monitor can show. I use this to identify and possibly edit problematic areas then switch back to normal.

There is another technique that works well with any monitor.

Make two copies of the image then convert one copy to printer space using RI, then back to your working color space also using RI.  All the colors that are in the printer's gamut will be quite close, within 1 dE and typically much less, but out of gamut colors will be exactly what is printed. Now take this image and put it in a layer with the other copy of the original. Select blending subtract mode. If there are no OOG colors you will get a black image. Add in a few percent of the original layer so you can just see the original. Colors that are OOG will become immediately apparent, lightening those OOG areas in proportion to how far OOG they are.

You can also use the Info panel to read out the LAB values. I set it to floating point which provides more precision in the LAB values. Just look at the same pixels in the original and the converted image to see what the printed LAB values would be.

I also have ColorThink Pro. It's useful for exploring profile gamuts and has lots of attractive graph options but it's quite slow at processing the pixels in a large image. Most people downsample by a large factor to reduce the processing time. However, even with that I don't find the program very useful as it's hard to relate colors shown in a 3D graph to those in the image. It's sort of like looking at a histogram v an image but harder to interpret. I have not found it as useful for the kind of work you are describing compared to the two processes using Photoshop.

[Brad P]

Brilliant Doug.  Thanks for those two suggestions.   I do have a wide gamut NEC monitor.  In particular I imagine the second suggestion will be most helpful in what I am trying to manage.  And I read your referenced posts yesterday on my journey to ColorThink Pro. 

I read that CTP can be slow, but haven't bought or used it yet.  I would be working with ~400MB files with a souped up 2009 Mac Pro (about as fast as a base model nowadays). I am willing to go have a coffee and wait a while if it could be helpful, but probably not lunch.  Are we talking coffee or lunch?

I get your comment that the histogram isn't helpful in managing nonclipped color detail to a printer profile.  In that light, if all CTP spits out are pretty three dimensional pictures without reference to specific color information that I can go back into and manage as I execute Photoshop commands or try to bring back into printer gamut out-of-gamut colors, then it seems worthless for what I am imagining it might assist with.   From a few videos on the Internet that I have seen, however, there appears to be quite specific color information that it might identify as colors to avoid blowing out or colors that can be better targeted to reign in.  Is that assumption wrong?  Is it more like looking at the histogram and trying to manage colors to a printer profile with that?

[Mark D Segal]

Brilliant Doug.  Thanks for those two suggestions.   I do have a wide gamut NEC monitor.  In particular I imagine the second suggestion will be most helpful in what I am trying to manage.  And I read your referenced posts yesterday on my journey to ColorThink Pro. 

I read that CTP can be slow, but haven't bought or used it.  I would be working with ~400MB files. I am willing to go have a coffee and wait a while if it could be helpful, but probably not lunch.  Are we talking coffee or lunch?

I get your comment that the histogram isn't helpful in managing nonclipped color detail to a printer profile.  In that light, if all CTP spits out are pretty three dimensional pictures without reference to specific color information that I can go back into and manage as I execute Photoshop commands or try to bring back into printer gamut out-of-gamut colors, then it seems worthless for what I am imagining it might assist with.   From a few videos on the Internet that I have seen, however, there appears to be quite specific color information that it might identify as colors to avoid blowing out or colors that can be better targeted to reign in.  Is that assumption wrong?  Is it more like looking at the histogram and trying to manage colors with that?

To use ColorThink Pro efficiently you would need to compress the image and use the "Unique Colors" option rather than All Colors. But the histogram is useful if used with the photo under softproof. In the final analysis, given the display you are using, with a correct colour management pipeline and reasonably accurate printer/paper profiling used under softproof, your eyesight provides the best guidance.

[Brad P]

Mark, I may be having a totally stupid Aha moment, now for the world to observe.  I've noticed the histogram in soft proofing changes dramatically, but never thought of using that in the sense of managing clipping against the printer gamut. 

Aha!

Thanks for that.  Now back to the pretty red and green flower picture that blew my mind last week. 

[Mark D Segal]

................... I've noticed the histogram in soft proofing changes dramatically, but never thought of using that in the sense of managing clipping against the printer gamut. 

Aha!

Thanks for that.  Now back to the pretty red and green flower picture that blew my mind last week.

Indeed, exactly what I suggested in reply #1 above. Depending on the paper you use, the change of the histogram will be more or less dramatic, with matte papers being more severe than luster/gloss papers, as to be expected given their generally narrower gamut.

[GWGill]

The best method I have found to further improve my post processing workflow may involve purchasing Chromix's ColorThink, analyzing individual files against my printer gamut as I go, and attempting to make more targeted adjustments.  Before I go to that route, is that about right or is there anything better?

Depends what you are after. Nothing is going to really beat hand rendering images from input to output color spaces, since you can make the subjective judgements that best please you. There are more automated ways of at least gamut compressing or clipping images though - for instance ArgyllCMS supports per-image gamut mapping workflows, but while there are a variety of controls and ways of using it, the automation means that there is less flexibility than doing it by hand, the results may therefore not always be quite what you want. So a combination of hand rendering and automated per-image gamut mapping may be the most efficient approach.

[Brad P]

Update:   I just purchased ColorThink (not Pro) and couldn't be happier.   Plotting a very difficult image (which BTW did not have to be downsized to load up promptly with a souped up Early 2009 Mac Pro) and comparing it against the printer colorspace, I was able to use Photoshop's selective color tool (the blacks slider only) together with luminosity masks generated with Tony Kuyper's luminosity curves to raise and lower the luminosity of fairly specific out of gamut colors, nudging each color up or down (with different layers) into the printer gamut, or at least close enough that I'm happy with the print.  It did meaningfully change the tone of the colors I manipulated, but not the hue or its saturation as other methods. With luminosity masks, these changes pretty seemlessly blended right in, appearing more like a subtle change in lighting conditions, not at all like the colors were being muted or shifted.  I'm confident this will work well with most out of gamut colors in problematic images, at least with large gamut papers. 

I haven't yet attempted to correct an image where the gamut of the image exceeds the outer limits of a printer gamut viewed from the top of the LAB chart down, and I imagine that will be more of a challenge (e.g, matte papers will likely have more of these problems).   But I'm confident ColorThink will give me at least more info to tackle that problem when it comes.  Knowing my luck, that will be tomorrow. 

As pointed out above, Photoshop's out of gamut warning is not very accurate. It was very easy to get the gray out gamut warnings to disappear, but loading up a flattened and saved file into ColorThink showed where a little more work needed to be done.   

Those of you who are familiar with the tools I mentioned above should be able to figure it out from what I've written.  If I can further clarify, let me know.  .