What's the point of using color spaces exceeding the visible spectrum?

[MarkM]

I have the design as an sRGB image in Photoshop and soft proof it using client's monitor color profile with RGB numbers preserved.
A monitor color profile can be used in the same way as any other device color profile like printers for example. Photoshop can display how the image will be reproduced on the other device. Soft proofing with RGB numbers preserved, displays how the color values will be reproduced with the other device, and without RGB numbers preserved, displays how the converted values will be reproduced with the other device.

This neither works in theory nor practice. I can attest to it not working in practice because I have two monitors side-by-side and just tried it. It's not even close—in fact the match is much worse when I soft proof to the monitor profile.

In theory it doesn't work either. The part you are overlooking is that the RGB numbers get pumped through a video card with it's own LUT calibration curves that are set when you calibrate. They are normally in the profile in the vcgt tag, but they are ignored when you softproof.

[jc1]

I was talking about soft proofing with Preserve RGB numbers.

My understanding is that you were viewing an sRGB image with a wide gamut monitor, and soft proofing with sRGB profile. This is a special case in that when Preserve RGB is on, the soft proof RGB value is the same as sRGB value.
 
If the soft proofing target is an printer profile, when Preserve RGB is on, the soft proofing is showing the actual RGB value sent to the printer and that is not the normal way we soft proof the printer output.

[digitaldog]

I have the design as an sRGB image in Photoshop and soft proof it using client's monitor color profile with RGB numbers preserved.

Having the clients profile, without the acutal display, and entire graphic system is useless. You have to have that client system to view what the client will see!

[MarkM]

Having the clients profile, without the acutal display, and entire graphic system is useless. You have to have that client system to view what the client will see!

While true, I can understand why people don't see this and why it's confusing. After all you don't need the client's printer and graphic system to softproof a printer—this is a normal and useful thing to do. I find it really difficult to explain to people why monitors (and projectors) are different.

[digitaldog]

While true, I can understand why people don't see this and why it's confusing. After all you don't need the client's printer and graphic system to softproof a printer—this is a normal and useful thing to do. I find it really difficult to explain to people why monitors (and projectors) are different.

Its different because a soft proof isn’t a print. If you want to see a print off the clients printer, you have to print it to that device. You could move that printer to another system and its a self contained process unlike a display system that uses a graphic system as part of the output process. Moving the printer and sending the same numbers will produce the same print from system A on system B. A soft proof isn’t the same thing. When you try to soft proof another display on your display, that’s where things get dicy.

[Peter_DL]

In theory it doesn't work either. The part you are overlooking is that the RGB numbers get pumped through a video card with it's own LUT calibration curves that are set when you calibrate. They are normally in the profile in the vcgt tag, but they are ignored when you softproof.

Hmm,
further thinking about it,
I’m wondering if Coloreason could just profile the client’s monitor without calibrating it,
so that the resulting monitor profile is only "descriptive" while the video card loading part is left out.

Peter

--

[Coloreason]

This neither works in theory nor practice. I can attest to it not working in practice because I have two monitors side-by-side and just tried it. It's not even close—in fact the match is much worse when I soft proof to the monitor profile.

In theory it doesn't work either. The part you are overlooking is that the RGB numbers get pumped through a video card with it's own LUT calibration curves that are set when you calibrate. They are normally in the profile in the vcgt tag, but they are ignored when you softproof.

Thanks for bringing this up, I understand what you are saying and it makes sense. We didn't think about that. However we did make a test with multiple monitors in our studio and believed that this will work based on the result. We used wide gamut monitors to soft proof standard gamut monitors. The monitors we have are calibrated so well with the hardware controls that when the LUT loads during start up we don't see any noticeable difference. However I've seen a noticeable difference when the LUT loads  in the past with other monitors and this may be the cases with our clients monitors which I didn't calibrate personally but another designer from my team. I'm glad we started to talk about this, I will reevaluate the reliability of our workflow.

[Coloreason]

Thinking about this, what happens in the following situation? Imagine an artist creating an artwork using a non-color managed program like one of the major 3D programs on a properly calibrated and profiled monitor but the artist dosent' have color managed programs like Photoshop. He sends his untagged artwork and his monitor profile to another person who uses another computer system with a color managed program, say Photoshop. Can the person with the color managed program recreate the colors as the 3D artist created them by assigning the received monitor profile to the received artwork?

[digitaldog]

Can the person with the color managed program recreate the colors as the 3D artist created them by assigning the received monitor profile to the received artwork?

In a color managed app, yes. But you could Assign and convert to a standardized color space like sRGB as well.

[Coloreason]

In a color managed app, yes. But you could Assign and convert to a standardized color space like sRGB as well.

Thanks, then how about if the monitor profile received from the 3D artist is assigned to a random image in Photoshop for the purpose to show how the 3D artist should see it if displayed on his monitor? I'm struggling to understand how is this different from doing it to an image received from the 3D artist.

[digitaldog]

Thanks, then how about if the monitor profile received from the 3D artist is assigned to a random image in Photoshop for the purpose to show how the 3D artist should see it if displayed on his monitor? I'm struggling to understand how is this different from doing it to an image received from the 3D artist.

I don’t understand the question. If the user builds an image in a non color managed app, but has a display profile, after opening that image in Photoshop and assigning the display profile, the color appearance shodul be maintained. The numbers have an assocaited scale. Then convert to sRGB and hand off.

[Tim Lookingbill]

Hmm,
further thinking about it,
I’m wondering if Coloreason could just profile the client’s monitor without calibrating it,
so that the resulting monitor profile is only "descriptive" while the video card loading part is left out.

Peter

Matrices that describe the colorants combined with the color temperature-(CT appearance varies between models and brands) work hand in glove with the vcgt RGB correction curves that maintain a linear distribution of 255 neutral gray tones between black and white.

All the slight bumbs, peaks and valleys you see in this RGB correction curve reduces and adds luminance to each 255 gray step to render a smooth gray gradient. Another display using only the matrices of the other display won't have these micro corrections that also affect the luminance of individual colors which acts on the preview similar to applying the same uneven curve to colors editing in Photoshop which I'm sure you've seen can affect the appearance of hue/saturation.

The matrices are only one leg of a three legged table of a display profile where not all colors are going to match between the two displays due to the other missing components that brings a display into a standard stable editing environment.

[Tim Lookingbill]

Below is a diagrammed explanation of why soft proofing with another client's display profile won't work. Displays are messy places to edit if not calibrated and profiled which attempt to clean things up a bit. Only using the matrice part of the client's display profile and leaving out the vcgt correction curves won't tell the whole story of how that image should look especially if that image was edited on a display that accounted for the vcgt loaded in the video card.

And we're not including gamut shape differences on top of that.

[Coloreason]

I don’t understand the question. ...

I was expecting that the answer to my question in post #67 would be, "no" based on what I've been told in this thread that monitor profile is useless on any other display than the one it was made for. But you answered yes and I'm trying to understand this. If person A creates an image A in a non-color managed program on computer/monitor A and person B using computer/monitor B can recreate the colors of the image A by assigning the color profile of monitor A to the untagged image  then I wonder how is this different than when a random image X created on computer/monitor B and assigned the color profile of monitor A cannot recreate how the colors will be displayed on computer/monitor A.

Let me know if my question is still not clear, I'll try to say it in another way.

[Coloreason]

Only using the matrice part of the client's display profile and leaving out the vcgt correction curves won't tell the whole story of how that image should look especially if that image was edited on a display that accounted for the vcgt loaded in the video card.

And we're not including gamut shape differences on top of that.

If it is not possible to create a color space describing how colors are displayed on a monitor by using a colorimeter and profiling software, I must have been with much higher expectation about what color management can do. I wonder if this is some sort of technical limitation or a matter of decision (user demand).
I guess they have to make 2 kinds of monitor profiles for a monitor, one for calibration and use for the current computer system and one that can be used on any other computer systems.

[Tim Lookingbill]

If it is not possible to create a color space describing how colors are displayed on a monitor by using a colorimeter and profiling software, I must have been with much higher expectation about what color management can do. I wonder if this is some sort of technical limitation or a matter of decision (user demand).
I guess they have to make 2 kinds of monitor profiles for a monitor, one for calibration and use for the current computer system and one that can be used on any other computer systems.

Didn't the diagram I posted above explain the two parts of a display profile?

There's no reason to use another display profile to show what images are suppose to look like on another calibrated/profiled display. The image will have the embedded profile/color space to make the image match on both displays. That's color management.

If your client isn't viewing tagged images on a calibrated/profiled display then the image isn't going to match anyway no matter what you do. It's that simple.

Even if you were to assign another display's profile containing the vcgt curves to the image as a way to soft proof, those curves would make the image look horrible because they are correcting for the anomalies of the other display that aren't present on the current monitor being soft proofed on. No two displays are exactly alike with regards to their calibration/vcgt correction curves.

[Coloreason]

Didn't the diagram I posted above explain the two parts of a display profile?

I understand there are two parts and I think I understand what they do but I still can't wrap my head around as to why monitors can't have profiles like printers for example. you send certain RGB value and it is displayed in a certain way. Knowing this why can't a color profile be created that can simulate the result without having the device?

[digitaldog]

You are on system A using display A. Display profile from System B is useless in showing you what the user saw on that system (B).

Person on system A creates a document and EMBEDS display profile, the document maintains the color appearance seen on system A.

[Coloreason]

....
Person on system A creates a document and EMBEDS display profile, the document maintains the color appearance seen on system A.

To make sure I understand this. Is the following statement correct?
Person on system A creates a document and EMBEDS display profile, the document when displayed on any system maintains the color appearance as seen on system A

[digitaldog]

To make sure I understand this. Is the following statement correct?
Person on system A creates a document and EMBEDS display profile, the document when displayed on any system maintains the color appearance as seen on system A

No

Not necessarily. Depends on the application used on the other systems, the condition (calibration) of the other systems. In an ICC aware app, with proper calibration, they should appear the same (within reason).