Outta gamut warning confusion: PS vs LR4

[Etrsi_645]

If I understood correctly, I thought we had been told to not use "gamut warning" in various releases of PS? Now in LR4 it seems to be a prominent and desirable thing to do..  If so, what has fundamentally changed to make this so?

[Wayne Fox]

1 reason is you can actively modify your file to pull things back into gamut.  (Yes you can sort of do that with PS but here much easier.)

watch Michael's video, he demonstrates this.  Haven't watch Julianne's yet but she might touch on this as well.

[Marlyn]

The fundemental problem with gammut warning in general still remains though.

You can't tell how MUCH out of gammut something is.  It could be a little, it could be a LOT.

[Wayne Fox]

Yes, fundamental problem but also quite challenging to solve "visually".

If you move the saturation slider of a corresponding color in the HSL panel you can watch the out of gamut gradually come back and sort of judge how far out any area is by how far you have to reduce saturation.  too soon to say how useful that may be (only played with it a few minutes), although it looks promising.

[hjulenissen]

Yes, fundamental problem but also quite challenging to solve "visually".

If some part of the image has a color that cannot be reproduced by the printer and/or not by the display that you are using to do the editing, what do we expect from the software? Perhaps a "mouse-over" could give us a delta-e or something in perceptually meaningful numbers of an exact spot?

I watched the Julieanne Kost video where she suggested to use HSV saturation or hue to manipulate out-of-gamut clipped areas back into gamut. The thinking seems to be "rather slighly less accurate colors, but smooth gradations, keep local detail". If what we always end up doing is simply pulling the color representation back/rotating just enough to keep every large-ish patch within gamut ("soft clipping"), then perhaps a simplistic automated conversion would do?

-h

[digitaldog]

There are two out of gamut warnings in LR4, one useful, one less so.

The OOG warning on the left tells you how OOG the image is compared to your display. I think that is kind of useful, it sure illustrates the differences working in an sRGB-like display and a wide gamut display. It can also help you keep a firm hand on the Vibrance or Saturation slider from going overboard.

The other OOG warning on the right shows you what is out of gamut compared to the profile you load. So like Photoshop, you get a large, ugly overlay and as Marlyn has stated, it treats a color a tiny bit OOG and one way the heck out the same. When you examine a 3D map of the image gamut in something like ColorThink compared to the output profile, the story is far clearer. But of course that isn’t a tool many would use and it is kind of useless for editing an image.

The idea of the OOG overlay in Photoshop predates it’s use of ICC profiles in Photoshop 5 back in 1998. The idea at the time was, look, your RGB image has all these out of gamut colors compared to the CMYK conversion mode, take the sponge tool or Hue/Sat and make that go away so your image is ‘in gamut’. Talk about a crude tool and workflow!

I saw a video yesterday where the OOG overlay was shown in LR4 and HSL was used to treat this area so it went away. But I don’t really see the point. Our output profiles will do this when we convert to the newer, smaller color space. For LUT based profiles, we have options in rendering intent and a good Perceptual intent will take OOG and in gamut colors into consideration when making the adjustment so to speak. I suppose for a few, very problematic images, you might want to try the HLS or similar route but man, that is a slow way to work and I have to wonder what the net results will be after you convert (which you have to do anyway right?).

A user setting for dE of out of gamut would be a bit more useful. ColorThink provides this. But talk about complexity for little use. The user has to be able to scale this dE coloring. And I suspect seeing 2 or 3 different colors overlaid would look kind of confusing. In the end, soft proof the image with the output profile and intent you intend to use. Edit if necessary.

[Geraldo Garcia]

Andrew,

You are, of course, absolutely right about the final conversion for the output profile taking care of the OOG colors. It will happen by clipping (relative colorimetric) or by fitting the image gamut inside the output profile´s gamut (perceptual).

The problem is that some (very specific) images simply look bad with perceptual rendering intent or suffer from loss of texture and detail in some areas when relative colorimetric is used. I remember an image of a texture made of different tones of out of gamut green, with perceptual the image was terrible and with relative colorimetric the texture turned into a solid tone of green. The solution was to use relative colorimetric as 80% of the image was correctly handled that way and to manually adjust only the texture to bring it into gamut.

Sure it was a lot of work and a very specific case, but to do it I had to use "softproof" and "gamut warning" switching them on and off as I worked.

Gamut warning is good to point you where you should look closely and pay more attention during softproof.

Best regards. 

[digitaldog]

The problem is that some (very specific) images simply look bad with perceptual rendering intent or suffer from loss of texture and detail in some areas when relative colorimetric is used. I remember an image of a texture made of different tones of out of gamut green, with perceptual the image was terrible and with relative colorimetric the texture turned into a solid tone of green. The solution was to use relative colorimetric as 80% of the image was correctly handled that way and to manually adjust only the texture to bring it into gamut.

But wasn’t that apparent from the soft proof? IOW, what did the ugly overlay bring to the party?

[Farmer]

I agree with Andrew (and very nicely spelled out).  Whilst it might seem like we need more power, I'm not sure it's practical in application.  I have ColorThink installed here at work and use it often enough - it's interesting, useful, but not an editing tool nor would it be even if available from within LR or PS.

Unless you have something with massive amounts of warning-overlay on it, then softproofing is the real answer, and now it's here.

[bjanes]

There are two out of gamut warnings in LR4, one useful, one less so.

The OOG warning on the left tells you how OOG the image is compared to your display. I think that is kind of useful, it sure illustrates the differences working in an sRGB-like display and a wide gamut display. It can also help you keep a firm hand on the Vibrance or Saturation slider from going overboard.

That is indeed valuable, since softproofing according to the appearance of the proofed image on the screen is only valid when the color being softproofed is within the gamut of the monitor. Wide gamut monitors are helpful here, but they only cover the AdobeRGB gamut (more or less) and that space is not all that wide and most experts suggest using ProPhotoRGB.

The other OOG warning on the right shows you what is out of gamut compared to the profile you load. So like Photoshop, you get a large, ugly overlay and as Marlyn has stated, it treats a color a tiny bit OOG and one way the heck out the same. When you examine a 3D map of the image gamut in something like ColorThink compared to the output profile, the story is far clearer. But of course that isn’t a tool many would use and it is kind of useless for editing an image.

Julieanne Kost does use that method in her Lightroom 4 tutorial. A 3D gamut map is more useful, but in Colorthink it merely shows the out of gamut colors and does not indicate in what part of the image these colors are located. The OOG indacator is better than nothing and is the only tool currently available when the gamut of the image exceeds that of the monitor.

A user setting for dE of out of gamut would be a bit more useful. ColorThink provides this. But talk about complexity for little use. The user has to be able to scale this dE coloring. And I suspect seeing 2 or 3 different colors overlaid would look kind of confusing. In the end, soft proof the image with the output profile and intent you intend to use. Edit if necessary.

The soft proof is helpful when the monitor gamut is equal to or greater to the color being softproofed, but what do you do when the monitor is the limiting factor? Rather than merely showing out of gamut or not, a dE pseudocolor display such as that used by Gamutvision (Norman Koren) would be helpful, although the multiple colors could be confusing.

But wasn’t that apparent from the soft proof? IOW, what did the ugly overlay bring to the party?

The overlay is helpful when the gamut of the image exceeds that of the monitor and softproffing is not informative. Consider this image (originally in ProPhotoRGB but shown here in sRGB for web display) which contains colors outside of the gamut of both one's monitor and printer. The image is to demonstrate a concept and not artistic content. Here I am using the Epson 3880 with Premium Luster with relative colorimetric rendering and the NEC PA241W profiled with the Spectraview II and the NEC supplied colorimeter.



And the 3D Colorthink gamut plot:



Some of the high luminance yellows are within the gamut of the printer and out of gamut for the monitor. Softproofing would not be helpful here. Some lower luminance yellow-greens are outside the gamuts of both the monitor and the printer, and the same considerations would apply. In this out of gamut area, softprofing would be helpful when the gamut of the monitor exceeds that of the printer (even though the colors are outside of the monitor gamut).

Regards,

Bill

[digitaldog]

Julieanne Kost does use that method in her Lightroom 4 tutorial.

But this is effective in what way when you’ll get all this done for you with the conversion of the profile? As I said, it may, (may) be useful in some very rare cases, I’d like to see some examples and wonder expect for those charging by the hour, who would do this.

The soft proof is helpful when the monitor gamut is equal to or greater to the color being softproofed, but what do you do when the monitor is the limiting factor?

The same thing you do when you deal with a source gamut which is larger than the destination gamut. You move on and print the image.

The overlay is helpful when the gamut of the image exceeds that of the monitor and softproffing is not informative.

Soft proofing is far from prefect but it is far better than having no soft proofing in predicting what the output will look like. But eventually you just print the damn image and look at the print. I don’t see how an ugly overlay that blocks your image on-screen helps here. Print the image.

[bjanes]

But this is effective in what way when you’ll get all this done for you with the conversion of the profile? As I said, it may, (may) be useful in some very rare cases, I’d like to see some examples and wonder expect for those charging by the hour, who would do this.
The same thing you do when you deal with a source gamut which is larger than the destination gamut. You move on and print the image.
 
Soft proofing is far from prefect but it is far better than having no soft proofing in predicting what the output will look like. But eventually you just print the damn image and look at the print. I don’t see how an ugly overlay that blocks your image on-screen helps here. Print the image.

The pseudocolor overlay could be useful in detecting very large gamut mismatches that could affect the visual appearance of the print. For example, a dE of 20 to 30 could be significant (as shown in the Gamutvision proof that I posted), whereas an out of gamut of several dEs might well not be significant. In my limited experience of soft-proofing with a wide gamut monitor, I find that trying to bring the whole image into gamut may be worse than simply allowing some colors to clip. When important details start to be lost by clipping, one should cut back on saturation. Or, as Julieanne showed in her tutorial, one could shift the hue to bring the color into gamut. Or one could reduce the luminance, since RGB spaces resemble in inverted pyramid at high luminances (as shown by the slicer in Colorthink). As you say, at some point, one must simply print the damned image.

Regards,

Bill

[digitaldog]

In my limited experience of soft-proofing with a wide gamut monitor, I find that trying to bring the whole image into gamut may be worse than simply allowing some colors to clip. When important details start to be lost by clipping, one should cut back on saturation. Or, as Julieanne showed in her tutorial, one could shift the hue to bring the color into gamut. Or one could reduce the luminance, since RGB spaces resemble in inverted pyramid at high luminances (as shown by the slicer in Colorthink).

So if manually bringing the image into gamut rather than allowing the profile to clip (which we both know isn’t exatly what a Perceptual intent provides), why follow the Julieanne example? I also have to wonder what the subsequent and necessary conversion to the space being soft proofed will do on top of this edit. I’m willing to be proven wrong, certainly in some examples but I’d sure like to see where all this editing, based on an ugly overlay, while actually seeing a soft proof of what you’ll end up with is at all useful. The soft proof should show us the issue you point out that is quite valid: excess saturation where detail suffers.