[BigMackCam]

Given that a colour space is basically a triangle within which points for specific colours can be mapped, and that a wider colour space is effectively just a bigger triangle, what's actually happening when I view a ProPhoto file on a monitor like mine that can only display 98% of sRGB?

See my previous post above, Dave. It depends on the colormetric rendering intent of the display profile (or soft-proofing / export conversion) your software is using.

If that intent is "perceptual", any saturations in the image that lie outside your display profile gamut will be converted to the maximums at the edges of the gamut and all other saturations within the display gamut will be adjusted to less saturated values, in order to create an overall representation that is "perceptually" consistent. I guess you could consider it a form of tonal compression, though I believe it's more complicated than that.

If, however, the intent is "relative colormetric", any out-of-gamut saturations will be clipped to the maximum available, whilst all saturations within the gamut will be directly mapped without adjustment.

There are other colormetric rendering intents, but these are the two I'm familiar with - and, I believe, the two most common and popular.

I'm not sure if I've explained this clearly - my terminology may not be spot on, but at least I know what I mean If it doesn't make sense, let me know and I'll try to re-word my explanation

---

Last edited by BigMackCam; 02-13-2020 at 03:53 AM.

[Dartmoor Dave]

See my previous post above, Dave. It depends on the colormetric rendering intent of the display profile (or soft-proofing / export conversion) your software is using.

If that intent is "perceptual", any saturations in the image that lie outside your display profile gamut will be converted to the maximums at the edges of the gamut and all other saturations within the display gamut will be adjusted to less saturated values, in order to create an overall representation that is "perceptually" consistent. I guess you could consider it a form of tonal compression, though I believe it's more complicated than that.

If, however, the intent is "relative colormetric", any out-of-gamut saturations will be clipped to the maximum available, whilst all saturations within the gamut will be directly mapped without adjustment.

There are other colormetric rendering intents, but these are the two I'm familiar with - and, I believe, the two most common and popular.

I'm not sure if I've explained this clearly - my terminology may not be spot on, but at least I know what I mean If it doesn't make sense, let me know and I'll try to re-word my explanation

Thanks Mike. That's a very clear explanation and it seems to agree with what my eyes are telling me. When I select a "perceptual" rendering intent in Photoshop's "convert to profile" dialogue the image tonality visibly becomes flatter. And when I select "relative colormetric" a lot of the superiority of the ProPhoto rendering is preserved.


Edit: And today I'm going to be spending some time shopping online for a wider gamut monitor. The last few days have made me realise that money invested in your display is just as important in digital photography as money invested in your camera and lenses.

[BigMackCam]

When I select a "perceptual" rendering intent in Photoshop's "convert to profile" dialogue the image tonality visibly becomes flatter. And when I select "relative colormetric" a lot of the superiority of the ProPhoto rendering is preserved.

Indeed so

The advantage with perceptual rendering is that the majority of the image's tonal differences are represented, even for a destination with considerably smaller gamut than working colour space, albeit through tonal compression (for want of a better term... I'm sure there must be one ). The disadvantage is that if your image contains a lot of heavily saturated tones outside the destination gamut, the flatter it can appear overall.

The advantage with relative colormetric rendering is that the accuracy of saturations that lie within the destination gamut is maintained... but, since saturations outside that gamut are effectively pegged to the boundary values, the disadvantage is that detail outside that gamut is compromised. There is, it would seem, no free lunch

EDIT: I've been doing some more reading myself this morning, and have come to the conclusion (for now, at least) that rendering intent at the time of export might best be chosen on an image-by-image basis, using soft-proofing and out-of-gamut warnings as a reference. If, with relative colormetric rendering intent, the image has no (or very few) out of gamut areas, it would seem the better choice to me, since all or most saturations will be accurately mapped to the output gamut - and therefore, the result should be close(r) to "WYSIWYG". If, however, there are many out-of-gamut areas and those areas contain important detail, then switching to perceptual rendering might be the better option, albeit at a cost...

Edit: And today I'm going to be spending some time shopping online for a wider gamut monitor. The last few days have made me realise that money invested in your display is just as important in digital photography as money invested in your camera and lenses.

That's certainly my view, although I know folks place differing importance on it. For me, a good quality, factory calibrated monitor with 100% sRGB - coupled with my existing Colormunki Display colorimeter and DisplayCAL software - was a decent compromise on gamut and colour accuracy vs cost. In retrospect, I wish I'd bought a model with 100% AdobeRGB coverage and built in calibration - but for my purposes, the monitor I have works very well...

[biz-engineer]

I believe the "elephant in the room" here is the "colormetric rendering intent" in each of the multiple conversions taking place

The problem is that we don't know what rendering intent is selected by the CMM or graphic card for the color matching. I've tried to figure that out by doing two exports: 1 export with "relative", and 1 export with "perceptual", then display both, I couldn't conclude anything. Maybe at the time I was still confused. Now I could think of another experiment to figure out what are the defaults used by the system when no rendering intent is user selected.

[BigMackCam]

The problem is that we don't know what rendering intent is selected by the CMM or graphic card for the color matching. I've tried to figure that out by doing two exports: 1 export with "relative", and 1 export with "perceptual", then display both, I couldn't conclude anything. Maybe at the time I was still confused. Now I could think of another experiment to figure out what are the defaults used by the system when no rendering intent is user selected.

In fact, we do. It's chosen at the time you create your .icc display profile... or, at least, it is in DisplayCAL (also in my HP laptop's proprietary profiling software)...

Also, consider that when exporting your image to a file with a defined profile such as sRGB or AdobeRGB and a specific rendering intent - perceptual, relative colormetric, etc. - that intent has nothing to do with the CMM, graphics card or display - it's only relevant to the colour space for the export file. When you display that file, how it is displayed will depend on the application, CMM and the icc profile for your monitor (which, may itself, be perceptual, relative colormetric or other)...

[biz-engineer]

When you display that file, how it is displayed will depend on the application, CMM and the icc profile for your monitor (which, may itself, be perceptual, relative colormetric or other)...

If an image file was exported as ProPhotoRGB, when it is displayed by a monitor having an sRGB type capability, obviously a rendering intent is chosen and we don't know what it is. When I display an image there is no popup window asking me is the rendering intent should be Perceptual or Relative colorimetric, this part is out of control and it has an impact on how an image is rendered. Now if I use GIMP, I can assign a color profile to an image and decide what the rendering intent is being used for color space conversion, I have full control. What I means is that the colors on display may look better when displaying a Prophoto image because we don't know how the display process work with the original Prophoto image to make it look good on a lesser sRGB display. Whereas when we do export an image to sRGB we do select a rendering intent which may no be as good as when performed by the display automation.

[BigMackCam]

If an image file was exported as ProPhotoRGB, when it is displayed by a monitor having an sRGB type capability, obviously a rendering intent is chosen and we don't know what it is. When I display an image there is no popup window asking me is the rendering intent should be Perceptual or Relative colorimetric, this part is out of control and it has an impact on how an image is rendered.

If you use a colour-managed application to view your exported images (as you should), you can usually select which profile it will use (typically the one you created for your display), or in many cases it will simply default to whichever one the OS is already using. For example, I use Firefox as my browser, and I explicitly configure it to have colour management enabled using my chosen display profile. That display profile was created with DisplayCAL, which allows me to choose rendering intent at the time of profiling. For my purposes, the display profile I use is set for perceptual rendering.

---

Last edited by BigMackCam; 02-13-2020 at 05:15 AM.

[biz-engineer]

If you use a colour-managed application to view your images (as you should), you can usually select which profile to use (typically the one you created for your display), or in many cases it will simply default to whichever one the OS is using. For example, I use Firefox as my browser, and I explicitly configure it to have colour management enabled using my chosen display profile. That display profile was created with DisplayCAL, which allows me to choose rendering intent at the time of profiling.

Sure. I have the same situation. But the rendering intent and color space (.icc or .icm) are two different thing. Below menu color management "enable" there is only the choice of .icc profile (e.g monitor profile), but nowhere there is an option of rendering intent. Did you mean that the .icc or .icm file itself contain the information of what rendering intent should be used by the CMM?

---------- Post added 13-02-20 at 13:24 ----------

I've found some info about Window Color System (WCS) here: Rendering Intents | Microsoft Docs
According to that web page, perceptual would be the default rendering intent for windows to display image files.

[BigMackCam]

Sure. I have the same situation. But the rendering intent and color space (.icc or .icm) are two different thing. Below menu color management "enable" there is only the choice of .icc profile (e.g monitor profile), but nowhere there is an option of rendering intent. Did you mean that the .icc or .icm file itself contain the information of what rendering intent should be used by the CMM?

Exactly that. The .icc or .icm profile includes rendering intent, and is set by the user in the profiling software (if available... it is in DisplayCAL, for example).

[Dartmoor Dave]

Exactly that. The .icc or .icm profile includes rendering intent, and is set by the user in the profiling software (if available... it is in DisplayCAL, for example).

Yep, DisplayCAL definitely gives you a choice of rendering intent (although my understanding is that apps can override that at will). I originally just used the default settings, which turns out to have been perceptual, but this afternoon (after reading around in the DisplayCAL forum) I've recalibrated using absolute colorimetric. I'd rather see all the colours within the range that my monitor is able to display exactly as they are, without any finagling to try to simulate colours from outside that range.

[BigMackCam]

Yep, DisplayCAL definitely gives you a choice of rendering intent (although my understanding is that apps can override that at will). I originally just used the default settings, which turns out to have been perceptual, but this afternoon (after reading around in the DisplayCAL forum) I've recalibrated using absolute colorimetric. I'd rather see all the colours within the range that my monitor is able to display exactly as they are, without any finagling to try to simulate colours from outside that range.

Interesting. I've been reading more since your post, Dave, since absolute colormetric rendering (something I'd not read up on before now) sounded tempting. Cambridge in Color has this to say:

Absolute colorimetric preserves the white point, while relative colorimetric actually displaces the colors so that the old white point aligns with the new one (while still retaining the colors' relative positions). The exact preservation of colors may sound appealing, however relative colorimetric adjusts the white point for a reason. Without this adjustment, absolute colorimetric results in unsightly image color shifts, and is thus rarely of interest to photographers.

This color shift results because the white point of the color space usually needs to align with that of the light source or paper tint used. If one were printing to a color space for paper with a bluish tint, absolute colorimetric would ignore this tint change. Relative colorimetric would compensate colors to account for the fact that the whitest and lightest point has a tint of blue.

The full article is here: Color Management: Color Space Conversion

I'm currently using perceptual rendering for my display profile, but I'm thinking I should be using relative colormetric... Further down the rabbit hole we go...

[Dartmoor Dave]

Interesting. I've been reading more since your post, Dave, since absolute colormetric rendering (something I'd not read up on before now) sounded tempting. Cambridge in Color has this to say:

My understanding -- and I could well be wrong about this because I'm new to DisplayCAL -- is that relative colorimetric assumes that your monitor is already set to a specific white point, then it twists the colour space around that intended white point value. So all colours will be skewed slightly away from absolute accuracy, in order to fit the monitor's predetermined white point. Absolute colorimetric calibrates a white point based on the colorimeter's measurements only and twists the colour space around that, ignoring whatever white value the monitor itself is set to. So all colours should be as accurate as possible, including white, as long as the monitor wasn't so far away from true white before calibration that the calibrated white point ends up out of gamut.

To put it more simply: relative colorimetric starts out by flashing up a white patch and assumes that the monitor is set so that this is what you want white to be. Then it calibrates all the other colours relative to that, which is fine as long as the monitor really was already set to what you wanted white to be before you started. Absolute colorimetric starts out by flashing up a white patch and assumes that the monitor isn't already set to what you want white to be. So it adjusts that starting white patch so that it is true white, and then it calibrates all the other colours around that measured true white.

At least that's the way I've interpreted it, although any sort of reliable documentation seems about as rare as hen's teeth and it might actually be exactly the other way round. I'll probably do a relative colorimetric calibration too, then switch between the two to find out which one gives me the best subjective results.

[biz-engineer]

Exactly that. The .icc or .icm profile includes rendering intent, and is set by the user in the profiling software (if available... it is in DisplayCAL, for example).

Spyder5 doesn't do that. And after reading microsoft documentation, I'm not sure that rendering intent is set in icc profiles. From what I read in WCS (Windows Color System) documentation, the rendering intent is set automatically via CMM settings. In windows, it is possible to open device settings, color management, go to advanced and modify the rules used by CMM to select the rendering intent depending of document type (photograph, pdf etc.). WCS documentation says that Perceptual is the default rendering for photographs when no other input is provided. It is said that Adobe CM engine called ACE has its own rendering intent selection rules. So I guess the same goes for Silkypix and any other photo viewer software when the rendering intent option isn't provided to us users.

[Dartmoor Dave]

Do systems like Spyder or DisplayCAL just create ICC profiles that Windows or other operating systems use as the basis for their own colour management? Or do the profile loaders that Spyder and DisplayCAL run at startup bypass the operating system's colour management and take direct control of it themselves? For example I use a plain white desktop background and both the Spyder and DisplayCAL systems alter the shade of that white, even though my understanding is that Windows itself doesn't colour manage the desktop.

Again, sorry if I'm putting things in an overly simplistic way. I feel I should keep adding caveats that this is a learning process for me, and to an extent my posts in this thread are a way of sort of thinking aloud and of course to seek very helpful input from others. There's no intention to cause any confusion.

[Dartmoor Dave]

I've now compared three different DisplayCAL rendering intents on both the desktop PC that I use for photo editing and the media centre PC in my living room. Subjective impressions:

Perceptual: The flattest looking colours and subjectively least accurate whites, with a slightly cool balance.

Relative Colorimetric: Subjectively very accurate looking whites and by far the most vibrant colours of the three, but to my eyes the colours seem oversaturated.

Absolute Colorimetric: The whites look subjectively very accurate and identical to the relative colorimetric whites as far as I can tell. Colours are more vibrant than the perceptual rendering, but not as heavily saturated as relative colorimetric. To my eyes, in particular looking at my own photos of the landscape in my area that I'm familiar with, the absolute colorimetric rendering seems the most accurate.

Those are just completely subjective impressions of course. It's quite surprising that using the same colorimeter and the same software can produce such visibly different profiles for the same monitor just by changing the rendering intent, but there it is. My overall conclusion is that I'm going to use the absolute colorimetric profile. . . and that monitor calibration in general is a flippin' nightmare.