A techique for lessening one drawback of wide-gamut monitors

[DanK]

I do my photo editing on a wide-gamut monitor, a NEC PA271Q. I follow the advice from Andrew Rodney and others to edit in the widest gamut the monitor can produce and deal with smaller gamuts at the end, when printing or posting to the web.

Recently I posted a close-up of a red gerbera (Center of unidentified red flower). I liked the shot, but when posted in sRGB, the reds were far off what I had on my monitor, and some areas showed the smooth loss of detail that one can get with oversaturation. No surprise. I figured it was just the small sRGB gamut.

But then I went to prepare it for printing on a relatively wide-gamut paper (Canson Baryta Photographique), and softproofing showed somewhat similar problems, which I wasn't able to correct acceptably. The reds I had produced were well out of gamut for the printer/paper combination. The major culprit, but not the only one, was strong selective color adjustments, one global and one local, in which I pulled cyan way down in the reds.

So, I started from scratch. I did copy some layers to a new copy, but most I did the editing again from scratch. I did all of the editing with softproofing turned on for that same paper, toggling back and forth initially between relative and perceptual colorimetric rendering until I decided on one for this image. The resulting image still looks much better with softproofing turned off, but this way I got an image that I think is acceptable for printing.

I'm wondering whether other people do something similar. It wouldn't be necessary for most images, of course.

[xpatUSA]

I do my photo editing on a wide-gamut monitor, a NEC PA271Q. I follow the advice from Andrew Rodney and others to edit in the widest gamut the monitor can produce and deal with smaller gamuts at the end, when printing or posting to the web.

Dan, I do similar for good stuff using ProPhoto 16-bit in RawTherapee. But I use sRGB working color space in the raw converter so that I can see potential gamut problems up front.

Recently I posted a close-up of a red gerbera (Center of unidentified red flower). I liked the shot, but when posted in sRGB, the reds were far off what I had on my monitor, and some areas showed the smooth loss of detail that one can get with oversaturation. No surprise. I figured it was just the small sRGB gamut.

Out of interest I look at your posted red gerbera center shot. The profile was tagged 'perceptual' but being a matrix type, my system of course rendered it as relative colorimetric. Sure enough, plenty of 3D gamut clipping in the reds:



And in the red channel (GIMP), that smoothing is quite visible in several places, especially at right:

But then I went to prepare it for printing on a relatively wide-gamut paper (Canson Baryta Photographique), and softproofing showed somewhat similar problems, which I wasn't able to correct acceptably. The reds I had produced were well out of gamut for the printer/paper combination. The major culprit, but not the only one, was strong selective color adjustments, one global and one local, in which I pulled cyan way down in the reds.

So, I started from scratch. I did copy some layers to a new copy, but most I did the editing again from scratch. I did all of the editing with softproofing turned on for that same paper, toggling back and forth initially between relative and perceptual colorimetric rendering until I decided on one for this image. The resulting image still looks much better with soft-proofing turned off, but this way I got an image that I think is acceptable for printing.

I'm wondering whether other people do something similar. It wouldn't be necessary for most images, of course.

At least when you use a printing profile, perceptual is what you get.

[Manfred M]

Dan it has little to do with wide gamut screens and a lot to do with the limitations (gamut) of the inks and papers.

I've discussed this with several very experienced printers and they all say the same things; reds are the most challenging colours to print. They suggested that one tries both perceptual and relative colorimetric rendering intents to see which one gives you the result that is more to your liking. One photographer told me that perceptual will tend to give you a print that looks more like what you see on your screen and relative colorimetric will give you a print that looks more like what the scene looked like in "real life".

The other issue is your paper; you like no-OBA papers and these tend to be yellower than ones with OBAs. When printing saturated colours, papers (especially lustre papers) with higher OBA levels will often give you a more pleasing result.

[DanK]

Ted,

Where did you get the 3D plot? I don't see that option in Raw Therapee.

If I were just posting on the web, I would leave the raw converter and everything else in ProPhoto to minimize the risk of processing artifacts but would use an sRGB display. That would be WYSIWG.

The problem is that when printing, gamut depends on the specific combination of printer and paper. I generally use coated papers, which give a wider gamut than sRGB, but gamut problems still arise. It probably varies depending on what you photograph, but in my work, the most common case of photos that go way out of gamut are flowers with intense reds. In most other cases, the OOG colors are minor enough, if present at all, that a few little tweaks are enough to provide an acceptable result.

BTW, I often do tonality edits of intensely red flowers with luminosity blending, effectively the same as doing the adjustments in the L channel of the Lab workspace, to avoid inadvertently boosting contrast. That wasn't enough in this case.

[xpatUSA]

Ted,

Where did you get the 3D plot?

I bought an app called ColorThink. I have V2.2.3.

I don't see that option in Raw Therapee.

If I were just posting on the web, I would leave the raw converter and everything else in ProPhoto to minimize the risk of processing artifacts but would use an sRGB display. That would be WYSIWG.

No it's not in RT.

Yes, the Sigma converter opens the raw in ROMM RGB space and all adjustment is done to that working file but the review and the color-picker values conform to the selected working space. Although the review image appearance stays the same, the color-picker values change depending on the working space selected.

I've noticed a similar effect in PS Elements where picker values change between "assigned" and "converted".

The problem is that when printing, gamut depends on the specific combination of printer and paper. I generally use coated papers, which give a wider gamut than sRGB, but gamut problems still arise. It probably varies depending on what you photograph, but in my work, the most common case of photos that go way out of gamut are flowers with intense reds. In most other cases, the OOG colors are minor enough, if present at all, that a few little tweaks are enough to provide an acceptable result.

BTW, I often do tonality edits of intensely red flowers with luminosity blending, effectively the same as doing the adjustments in the L channel of the Lab workspace, to avoid inadvertently boosting contrast. That wasn't enough in this case.

Understood. Stay well ...

[DanK]

Dan it has little to do with wide gamut screens and a lot to do with the limitations (gamut) of the inks and papers.

Manfred, I must have been unclear. What I was describing wasn't a printing problem per se. It is the fact that in some cases, the discrepancy between what a screen and a printer can produce is greater when one is using a wide-gamut display set to its full gamut. The point of my post is that sometimes, this discrepancy is so large that it can't be overcome by post-hoc edits during softproofing. In other words, having a wide-gamut monitor can lead one to make edits that go well beyond what printers can produce. I went farther astray in this case than I would have had I been using an sRGB monitor because I was happily creating gorgeous reds that happened to be far out of gamut for my printer/paper combination. My solution is to start again and turn soft proofing on at the outset. This rarely happens with standard narrow-gamut monitors.

They suggested that one tries both perceptual and relative colorimetric rendering intents to see which one gives you the result that is more to your liking.

Of course. That's the first thing to do if you have a serious issue with out-of-gamut colors. In this case, the problem was far too severe to resolve this way. However, as I pointed out, in my second round, with softproofing turned on at the outset, I toggled between these two rendering intents to see which did the least damage.

The other issue is your paper; you like no-OBA papers and these tend to be yellower than ones with OBAs. When printing saturated colours, papers (especially lustre papers) with higher OBA levels will often give you a more pleasing result.

I was preparing to print on Canson Baryta Photographique (the original, of which I still have a supply). As you know, that paper has OBAs, and it has a very wide gamut. However, that's an interesting point. I can switch profiles among papers to see how much difference that makes. I suspect not much, given that it is blues that OBAs enhance, but I don't know for a fact. The Canson paper is low-OBA, but I have profiles loaded for high-OBA papers (e.g., some cold white Red River papers and Canson Prestige) and some zero-OBA papers (like Canson Platine and Breathing Color River Stone Rag).

That does raise an interesting question. I wasn't looking at prints. I was looking at soft-proofed images on screen. As you know, papers with OBAs look quite different under natural light with UV than under most artificial lighting because it's UV light that causes OBAs to fluoresce. I wonder which kind of light is assumed in softproofing algorithms.

Yes, intense reds can be extremely difficult. I have spent countless hours on images of intensely red flowers over the years, and I haven't always been successful. Almost all of the gamut-related failures I have had have been because of reds. It was actually oversaturation in an image of an intensely red tulip years ago that led me to start doing some tonality adjustments in luminosity blending mode. I also worked in Lab mode a bit for the same reason, but I found that cumbersome, and as I showed in a thread we had some years ago, one can get essentially identical results by applying a luminosity blend mode to tonality adjustments.

[xpatUSA]

... Yes, intense reds can be extremely difficult. I have spent countless hours on images of intensely red flowers over the years, and I haven't always been successful. Almost all of the gamut-related failures I have had have been because of reds.

Indeed, reds get the blame most of the time although I have encountered the occasional sunflower that fooled ACR 4.5.

I wonder where "intense reds" lie on the good old L*u'v' CIE diagram and perhaps if they are within the Gospel according to Pointer:



https://www.tftcentral.co.uk/article...#_Toc379132054

Or, I wonder what the dominant wavelength of "intense reds" is?

Anyone?

P.S. lots and lots and lots of info here:

https://philservice.typepad.com/Poin...splays_v2.pdf4
.

[xpatUSA]

I wonder where "intense reds" lie on the good old L*u'v' CIE diagram and perhaps if they are within the Gospel according to Pointer:



Or, I wonder what the dominant wavelength of "intense reds" is?

Here's Dan's in the same CIE space (Luv):

[DanK]

Fascinating. Thanks for posting this. I need to read these more carefully to be halfway confident that I understand them, but to the extent that I do: some of the colors in question have to be within Pointer's gamut but outside of both sRGB and the printer's gamut, I think, because after all, I could see them on my wide-gamut monitor but couldn't reproduce them elsewhere.

Interestingly, one of my main edits the first time around must moved the hue along the line of purple toward the monocromatic reds on the edge because I was pulling out cyan from the reds. And if I recall--I'm at a different computer now--when I simply desaturated without pulling down cyan, the color looked more purplish. Pulling out the cyans without desaturating produced a seemingly more nearly monochromatic read on the wide-gamut monitor. But when I displayed the image in sRGB and in softproofing, it didn't become visibly purpler; it became if anything a tad more orange. This must have something to do with how the ICC causes the software to map to the smaller space. At least, this is how I remember it.

My first failure with this problem was a cardinal flower, years ago. I've had several more failures since, although I did with difficulty coax one flower that had out of gamut reds into something that prints fine and looks acceptable in the sRGB gamut:

[Manfred M]

Thanks for the clarification Dan.

I am somewhat suspicious of Soft Proofing; it is an emulation of a reflective light, CMYK, subtractive process using a transmitted light, RGB, additive process.

Depending on who you listen to, most people suggest that it is a 80% - 95% accurate / useful approach. None of the high end print makers I studied under use or recommend using it because of its limitations. They both are 100% committed to making test prints and basing that process, so it is not a tool that I have ever adapted to regular use. Nicely said, I am not 100% confident with what I see using that mode in Photoshop.

I have occasionally run into colour issues, especially with the reds and as Ted has noted, yellows can be problematic at times too. I remember struggling with some yellow hats that Buddhist monks were wearing. I am not necessarily looking for colour accuracy, but rather for realistic looking gradations throughout the image. What I found as worked a couple of times is to work in sRGB in my raw conversion, rather than going to a wider gamut colour space. ProPhoto can definitely be be problematic when pushed hard, even in 16-bit mode.

While I understand Andrew Rodney's approach (and he is not the only one recommending that workflow), sometimes things can go awry during the colour space conversion process. I found that I could not push too hard and that seems to preserve the subtle graduations better at times. It seems counter-intuitive, but in a few really problematic cases that worked well for me. I suspect my problematic cases are not as challenging as the colours in flowers.

When it comes to papers I find that high OBAs can have some positive impact on colours. OBA free papers tend to be a touch warm. I am looking at a sample of Canson Baryta Photograpique II and Hahnemühle Silk Baryta and the reds are more "realistic" on the "whiter", Hahnemühle paper. I suspect the warmer tones of the OBA free base is the underlying reason.

[DanK]

There's a lot to sort out here.

First, the issue in my OP wasn't tone; it was bad rendering of out of gamut colors (which I'll abbreviate OOG). I don't think that problem had anything to do with OBAs or the lack thereof. After reading your earlier post, I started searching online for information about OBAs and gamut, and I couldn't find anything with any technical backup, so I posted on photopxl in the hope that Mark Segal would reply. He did. You can read the exchange here: https://photopxl.com/forums/topic/ob...me/#post-30617.

Re soft proofing: I find it helpful to separate two questions. Is it useful? And is it sufficient? In my experience, the answer to the first is "absolutely," even though it often suggests only modest edits, e.g., slight changes in contrast or tone. The answer to the second is "sometimes". I find that with large-gamut papers I know well and images with no OOG colors, test prints rarely surprise me. Not never, but rarely. With difficult images, in particular, with images that have a lot of severely OOG colors, all bets are off. I would still start with soft proofing to make initial corrections, but I would want to rely on test prints. The Itoh peony image above went through a number of test prints.

Re tone: I too prefer cold white papers. that's why I used the original Canson Baryta Photographique for a number of years. I find the tone of that paper to be what I want in most cases, and given that the OBA level is low, I accepted that. Recently, as you know, when that paper was discontinued, you and I both went through the process of testing a substantial number of papers, and I wanted to stay away from high-OBA papers for archival reasons, not for taste. One of the ones I ended up with, Breathing Color River Stone Rag, is a zero-OBA paper, and it is slightly warmer in tone than the Canson paper. The difference, however, is very slight when I look at test prints. Moreover, it's slighter under some types of artificial light, as UV is what makes OBAs fluoresce. So I am still of two minds about this, but I must say that River Stone produces stunning images.

Re Rodney's approach: I disagree with him on one point, and the reason is precisely what you point out above: you can never really match an emissive display to a reflective one. I print in Lightroom, which has two print adjustments that do NOT show up in soft proofing: a contrast adjustment and a brightness adjustment. The way to use these is to compare test prints to screen displays. I found early on that when I thought I had a good match on screen, I would end up with prints that seemed very slightly lower in contrast and very slightly darker than I wanted, despite soft proofing, a proper monitor illumination, and so on. My translation from emissive to reflective is slightly off. By looking at test prints, I found that +10 on both scales got me what I wanted. Rodney considers this an unacceptable kludge--his word--because it can't be replicated on screen. But that's entirely the point, IMHO: it's a way of calibrating your printing for unintended differences between reflective prints and emissive soft proofs. I've only had to change those settings once, when a test print showed that the tiny contrast boost was creating problems in a print of a monochrome image.

[xpatUSA]

There's a lot to sort out here.

First, the issue in my OP wasn't tone; it was bad rendering of out of gamut colors (which I'll abbreviate OOG). ...

Reference my earlier gamut diagrams and the understanding thereof:

I opened a yellow flower raw in the converter, adjusted the brightness down a tad. The color picker in sRGB working space showed bottomed blues which, for yellow, indicate OOG. I saved it as a JPEG. Then, while still open, I made several adjustments more - mainly reducing saturation to bring up the blues then putting the histogram to look "OK". Here are the converter review images:



At right is the unadjusted output looking a little more pretty than the other. At left, slightly de-saturated, etc.

Adjustments were made in the converter so as not to export a gamut-clipped image for further processing.

Now the less easy-to-understand part. I opened each image's color gamut and the sRGB gamut in ColorThink. I selected the less-complicated 2D chart option and selected the CIE 1976 u'v' diagram and also zoomed into the green/yellow part so as to easily see the more relevant color points:



The dots are displayed in their own color (i.e. position on the diagram) but do note that they are also displayed in their original lightness i.e. darker equals less bright. Had the diagram been 3D, darker dots would have been separate and lower down.

Two things to note: Firstly the un-adjusted yellows are butted up against the sRGB gamut boundary - as expected from seeing a bottomed channel (blue) in the converter. Secondly the spread of yellows is more in the adjusted image at left. The degree of spreading in the direction of the white point (towards lower right) indicates the amount of "saturation contrast" if you will. More of such is arguably better than less, IMHO.

Interpreting gamut diagrams does take a lot of getting used to.

One more thing. ColorThink can examine images profiled as CMYK and one can add profiles for comparison. It even comes with a couple of SWOPs. I imagine that would be quite useful for those seriously interested in printing. I myself have two sRGBs - the common IEC one but also ICC's 2014 version.

So workflow is significant and -of course- mine is best, LOL.

[edit] Out of interest, I just extracted the blue channel from the embedded JPEG in the raw file and observed that 0.68MP were bottomed out of 4.6MP total! Not good ... in other words, Sigma's in-camera raw processing is nowhere near as good as mine.

[DanK]

Andrew Rodney has joined the discussion at the link I gave above. Mark Segal wrote that it seems possible that OBAs will increase gamut size in the near UV range. Rodney says it makes no sense to him that OBAs would affect gamut. Neither has comparisons between papers with and without OBAs--it would be hard in any event to partial out the impact of other differences among the papers--but Mark does have some empirical data from comparisons of different profiles.

Mark did give the gamut volume he obtained from the no-OBA River Stone paper, and it was very large.

So while I'll continue to dither about OBAs for other reasons, I think gamut won't be one of them.

[xpatUSA]

Fascinating. Thanks for posting this. I need to read these more carefully to be halfway confident that I understand them, but to the extent that I do: some of the colors in question have to be within Pointer's gamut but outside of both sRGB and the printer's gamut, I think, because after all, I could see them on my wide-gamut monitor but couldn't reproduce them elsewhere.

Interestingly, one of my main edits the first time around must moved the hue along the line of purple toward the monocromatic reds on the edge because I was pulling out cyan from the reds. And if I recall--I'm at a different computer now--when I simply desaturated without pulling down cyan, the color looked more purplish. Pulling out the cyans without desaturating produced a seemingly more nearly monochromatic read on the wide-gamut monitor. But when I displayed the image in sRGB and in softproofing, it didn't become visibly purpler; it became if anything a tad more orange. This must have something to do with how the ICC causes the software to map to the smaller space. At least, this is how I remember it.

My first failure with this problem was a cardinal flower, years ago. I've had several more failures since, although I did with difficulty coax one flower that had out of gamut reds into something that prints fine and looks acceptable in the sRGB gamut:

Here's the gamut of the above, Dan, compared with IEC sRGB and also a SWOP profile FWIW:



Hopefully of some interest although I realize that the profile is not for the paper you use ...

[DanK]

Ted,

Very interesting. I don't know how the SWOP gamut compares with the gamut of the Canson paper, but since I managed to get it to look OK in sRGB, this is informative. Correct me if I'm wrong, but the way I read this is:

1. Except for yellows, I have a large percentage of nearly-primary colors (near the right edge).
2. The piling up at the gamut boundary, particularly in the reds, indicates that I had a lot of OOG reds that were moved into the gamut by my imposition of the sRGB color space.

This suggests that some yellows were also OOG, but I didn't find it hard to make the yellows look reasonable, perhaps because a substantial portion of the yellows were not OOG. The real problem was the deep reds, which typically rendered as a brownish red in soft proofing.

I actually sent this one to a lab as well because I wanted it printed on metal to go over a fireplace, and I wanted it larger than I can print. All of the test images I ordered had brownish reds, so I gave up and printed it myself and framed it. it looks fine.

Dan

[xpatUSA]

Ted,

Very interesting. I don't know how the SWOP gamut compares with the gamut of the Canson paper, but since I managed to get it to look OK in sRGB, this is informative. Correct me if I'm wrong, but the way I read this is:

1. Except for yellows, I have a large percentage of nearly-primary colors (near the right edge).

Sounds about right, Dan, and the lightness of those points is indicating around the mid-tones as found in the central part of your image, the primary colors being right at the corners of the sRGB gamut.

2. The piling up at the gamut boundary, particularly in the reds, indicates that I had a lot of OOG reds that were moved into the gamut by my imposition of the sRGB color space.

Well, if the profile was truly perceptual (i.e. not simple matrix) then the reds would have moved proportionally into the sRGB gamut. If not, then the OOG reds would have been brought only to the edge of the gamut if the CiC tutorial is to be believed - see 'relative colorimetric'.

This suggests that some yellows were also OOG, but I didn't find it hard to make the yellows look reasonable, perhaps because a substantial portion of the yellows were not OOG.

Agreed.

The real problem was the deep reds, which typically rendered as a brownish red in soft proofing.

I actually sent this one to a lab as well because I wanted it printed on metal to go over a fireplace, and I wanted it larger than I can print. All of the test images I ordered had brownish reds, so I gave up and printed it myself and framed it. it looks fine.

I've found that pure reds look brown to me when their brightness is lowered. That is to say that brown is not actually a hue, unlike orange for example. I read that somewhere but don't have the link any more ...

... this one by a painter seems relevant:

https://swannportraits.com/brown-is-not-a-color-either/

I suppose that means that lightness and saturation (chroma?) are not related and that a really dull-looking red could indeed be highly-saturated.

And if you look at the sRGB gamut in 3D Lab, it is quite obvious that the gamut boundary reduces significantly as the Lightness gets lower. Which is why brightly-lit flower parts are less "saturated" than shadowed parts. I'll extract a channel in the GIMP and have a look at that.

Later,

[Manfred M]

I don't know how the SWOP gamut compares with the gamut of the Canson paper

The SWOP (Specification for Web Offset Publications) are not something most amateur photographers ever come across and are aimed at printing on traditional offset presses using CMYK inks. Something like the magazines you find in the grocery stores or the (soon to be discontinued) IKEA catalogue would be using the SWOP colour space. It is slightly wider than the sRGB colour space; if I recall correctly it does has a wider gamut in the green to blue range and yellows than sRGB and is narrower than sRGB in the reds and magentas.

The papers used in the SWOP specs can be coated and uncoated, but are usually nowhere near the quality of what we use in the fine art inkjet process.

I've only had to prepare images for this type of reproduction once (this was last year, so it's still fairly fresh in my mind) when some of my work was used in a magazine.

[xpatUSA]

... I suppose that means that lightness and saturation (chroma?) are not related and that a really dull-looking red could indeed be highly-saturated.

And if you look at the sRGB gamut in 3D Lab, it is quite obvious that the [gamut] boundary reduces significantly as the Lightness is less. Which is why brightly-lit flower parts are less "saturated" than shadowed parts. I'll extract a channel in the GIMP and have a look at that.

Because I've been talking about Lightness, I decomposed the peony into HSL.

Here's the Lightness layer:



And here's the Saturation layer:



In the Saturation layer , white is 100% saturation, black is none. Observe how the whites coincide mostly with the darker areas in the Lightness layer - and vice versa.

[DanK]

Thanks. This was something I hadn't understood.