ProPhoto to sRGB?

[smthopr]

I was recently editing a photo in ProPhoto working space on a wide gamut display calibrated to REC709 (I'm doing some video work).

I flattened the image and converted to sRGB space and noticed a color shift. The image is low saturation and should be completely in REC709 space. But there is a slight color shift upon conversion.

Why?

[digitaldog]

I flattened the image and converted to sRGB space and noticed a color shift. The image is low saturation and should be completely in REC709 space. But there is a slight color shift upon conversion.

We need low rez examples to examine. There should be no 'color shifts'. The color appearance could change slightly depending on the gamut of the two but not a shift.

[Doug Gray]

I was recently editing a photo in ProPhoto working space on a wide gamut display calibrated to REC709 (I'm doing some video work).

I flattened the image and converted to sRGB space and noticed a color shift. The image is low saturation and should be completely in REC709 space. But there is a slight color shift upon conversion.

Why?

When converting from a wide gamut to a narrow one you may get different colors because of clipping. Conversion is done by a matrix and when the RGB values that occur after conversion are negative they are clipped to 0. This can cause slight shifts hue and significant decreases in saturation but also increases luminance and often that is the largest effect.  The opposite can occur as well. When RGB levels are clipped at 255 luminance is decreased.

This effect is not observable when the monitor is in the same space as the conversion target since the clipping is the same for both. So, rather counterintuitively, folks that notice this are those that run wide gamut monitors. If one runs a profiled monitor at sRGB and converts to sRGB they will remain blissfully unaware of the changes.

What is printed from ProPhotoRGB is different. ProPhotoRGB is first converted to Lab and most, but not all, colors as ICCLAB limits a and b to values in the range of -128 to 127. This isn't the primary limitation as most realizable ProPhotoRGB colors can be defined in those limits. All , non-specular, reflected colors are within that space.

The problems with printing directly from ProPhotoRGB v first converting to sRGB (or any other matrix type space space) is that the path is different. After the color is converted to Lab, lookup tables map the color to what a printer can actually print and that can be a color that is outside or inside both sRGB and AdobeRGB. This process is completely different. Worse, the way colors are mapped when they are outside the printable gamut are undefined. Even for printable colors they are undefined using Perceptual Mode. It's basically up to the profile software peeps who try to create "pleasing" colors. At least it's within the context of a defined viewing environment for the more recent V4 profiles. It may, or may not match one's expectations.

[smthopr]

We need low rez examples to examine. There should be no 'color shifts'. The color appearance could change slightly depending on the gamut of the two but not a shift.

Here's the screen shot...  Look at the road in the lower right.  It's more red or magenta in the sRGB version

[digitaldog]

Here's the screen shot...  Look at the road in the lower right.  It's more red or magenta in the sRGB version

Look identical to me, hence the suggestion for the actual but low rez file. Oh well.

[smthopr]

Look identical to me, hence the suggestion for the actual but low rez file. Oh well.

They are not identical.  Open in photoshop and use the eye dropper to measure and you'll see.  To make it obvious, cut the image in half, and superimpose them as layers, and turn one off and on to see the difference.
Since I posted, I also noticed a change in overall gain of the image as well.  Sky is obviously lighter in the sRGB version and gives a different impression.

When I'm in ProPhoto space, photoshop is doing an "on the fly" conversion to my monitor space using relative colormetric intent,  yes?
But converting to sRGB, which is darned close to my monitor calibration, using relative colormetric intent, causes a different appearance.  Yes subtle, but different.  Why is that? I don't think this image challenges either gamut...  Or, do those little bright red tail lights cause some shifting of all the colors in the remapping to sRGB?

[Lundberg02]

I think you'll find that the conversion to your monitor space is perceptual.

[D Fosse]

If different conversions look different on screen, it's the display profile. ProPhoto to monitor is a different calculation than sRGB to monitor.

Recalibrate, and leave the monitor at native primaries and whatever white point you have it set to.

[digitaldog]

They are not identical.  Open in photoshop and use the eye dropper to measure and you'll see. 

I'd certainly hope they are not identical! Converting from ProPhoto RGB to sRGB should produce differing values. But what's the point? I asked for data to test and you're providing screen captures. IF and when you're ready for assistance, follow the instructions and I'll try to help you. Otherwise, I'm done.

[smthopr]

I'd certainly hope they are not identical! Converting from ProPhoto RGB to sRGB should produce differing values. But what's the point? I asked for data to test and you're providing screen captures. IF and when you're ready for assistance, follow the instructions and I'll try to help you. Otherwise, I'm done.

Sorry to bother you then...  I was just curious about the effect and I thought others here also would be.

[digitaldog]

Sorry to bother you then...  I was just curious about the effect and I thought others here also would be.

You got an answer in the 2nd post: There should be no 'color shifts'. The color appearance could change slightly depending on the gamut of the two but not a shift.

[Doug Gray]

I was recently editing a photo in ProPhoto working space on a wide gamut display calibrated to REC709 (I'm doing some video work).

I flattened the image and converted to sRGB space and noticed a color shift. The image is low saturation and should be completely in REC709 space. But there is a slight color shift upon conversion.

Why?

I looked at the image and attached profile. Attached profile is a standard Eizo s/w one with a matrix.
The screen grabs have the monitor's profile attached. There are a lot of jpg compression artifacts that makes it difficult to compare pixel to pixel. Screen captures should not be compressed to evaluate subtle differences.

There clearly is a difference, though small, in the captures. Even if the original was 8 bit the differences are too large to explain.
The differences are not due to monitor profile issues unless one screen capture was interpreted as sRGB. The attached image is tagged with the monitor's CX271.... profile.
I created a dark gray image in ProPhotoRGB then added random color noise but within smaller gamuts. After converting a duplicate to sRGB and comparing the ProPhoto and sRGB overlays in photoshop there was no visible difference at all and I am also using Eizo.

It's not a gamut or monitor profile issue per se but some sort of process issue possibly in the "flattening" step which could well be shifting colors.

[D Fosse]

The differences are not due to monitor profile issues unless one screen capture was interpreted as sRGB

They can if the profile is somehow defective.

But actually, on closer inspection, I suspect that what we see here is the well-known Photoshop bug that causes heavy red/cyan banding in the shadows in ProPhoto files. It can take different forms with different display profiles, but never goes entirely away. It happens even with sRGB/Adobe RGB as display profile, so there's no doubt the bug is real.

This color banding happens only in (16 bit) ProPhoto files, with GPU set to "Normal" or "Advanced" modes. In "Basic" mode, the display color management logic is shifted back from the GPU to the CPU, and the banding disappears. ARGB and sRGB are not affected.

This bug has been well documented on the Adobe Photoshop forum from 2012 on. Chris Cox said they'd "look into it", but nothing's happened.

So try to set GPU to basic and check again.

[Doug Gray]

They can if the profile is somehow defective.

But actually, on closer inspection, I suspect that what we see here is the well-known Photoshop bug that causes heavy red/cyan banding in the shadows in ProPhoto files. It can take different forms with different display profiles, but never goes entirely away. It happens even with sRGB/Adobe RGB as display profile, so there's no doubt the bug is real.

This color banding happens only in (16 bit) ProPhoto files, with GPU set to "Normal" or "Advanced" modes. In "Basic" mode, the display color management logic is shifted back from the GPU to the CPU, and the banding disappears. ARGB and sRGB are not affected.

This bug has been well documented on the Adobe Photoshop forum from 2012 on. Chris Cox said they'd "look into it", but nothing's happened.

So try to set GPU to basic and check again.

I believe you nailed it. Not a profile or gamut issue but an Adobe bug.  I've also seen this in proofing 16 bit LAB files. Have had GPU turned off for some time now.

[digitaldog]

This color banding happens only in (16 bit) ProPhoto files, with GPU set to "Normal" or "Advanced" modes. In "Basic" mode, the display color management logic is shifted back from the GPU to the CPU, and the banding disappears. ARGB and sRGB are not affected.

I see no differences with any of those settings on this end.

[D Fosse]

It is dependent on the display profile to a certain extent. Lots of people chimed in to confirm when it was discussed on the Adobe forum, but people saw it differently. Most saw cyan banding, some cyan/red, and others just luminance banding.

So perhaps some configurations go entirely clear, I don't know.

I did see it on an NEC P232/SV II, but IIRC only as luminance banding. Matrix is usually better than LUT.

In any case, I believe it's the problem here.

[digitaldog]

So perhaps some configurations go entirely clear, I don't know.

NEC PA272W. I might need to see what shows up on the MacBook Pro screen but since the two are so vastly different and I don't care all that much what 'critical' color looks like on the Retina, not a big deal. Maybe I need to relaunch PS after each setting? Tried it once, saw no difference. ProPhoto gradient from black to white looks neutral. Running latest version of Photoshop CC.

[digitaldog]

Don't see it on the Retina display either. Gradient isn't anywhere as smooth as the SpectraView as expected. But it appears neutral. Maybe the latest version of CC fixed the bug?
I wonder if Dither when creating the gradient plays a role?

[D Fosse]

Quick testing it again now (Eizo/Colornavigator), it seems OK here too...maybe it is fixed? It would be about time. I need to look more closely at this (I currently only have matrix profiles).

But to recap - the interesting thing wasn't so much the banding you did see in these ProPhoto files, but the fact that it disappeared immediately when switching to "Basic" mode or GPU off. So that was instant confirmation and you could pretty much rule out any other causes.

I've also seen other issues than banding with this, most notably black clipping to level 5 or 6 in PP. That's also visible in the examples I posted above. Again the problem went away in Basic mode or GPU off.

[digitaldog]

But to recap - the interesting thing wasn't so much the banding you did see in these ProPhoto files, but the fact that it disappeared immediately when switching to "Basic" mode or GPU off. So that was instant confirmation and you could pretty much rule out any other causes.

GPU bugs are nothing new for Adobe  . I hope indeed the Photoshop bug you reported is now fixed.