Understanding tonemapping Prophoto-RGB vs Adobe-RGB

[kers]

I just printed two 1-255 grayscale images

one is Adobe-RGB  1-255
the other Prophoto-RGB 1-255

I printed them both the same:  relative colorimetric with blackpoint compensation using the same profile.

Result is that the dark tones are lighter on the Prophoto-RGB than on the Adobe-RGB.

I would have expected that there would be no difference. ( the light tones are almost similar)

Maybe someone here has a clue? When you read the Lab Values of both images it indicates the same: dark areas are lighter- you can see it also on screen.

Does this mean in general that you have lighter dark areas on a print when you work in Prophoto-RGB compared to a print from Adobe-RGB?

PS i just think it has to do with the larger gamut of Prophoto RGB vs adobe-RGB, making the steps larger.

[Doug Gray]

You are seeing the effect of different gamma encoding. Adobe RGB is 2.2, ProPhoto RGB is 1.8.

An example, RGB(128,128,128) has an L* value of 54.0 in Adobe RGB while it has 60.7 (significantly lighter) in ProPhoto RGB. This difference increases as the values get darker. With RGB values of (30,30,30) the differences expands and is quite noticeable at L* = 8 and 16.

When you convert from one matrix based colorspace to another, the values are changed such that the Lab values remain unchanged for all values within the gamut of the destination space. Neutrals are always in gamut between matrix spaces hence not changed upon conversion.

This is all purely a result of gamma differences. Note that sRGB is often stated, incorrectly, to have a gamma of 2.2. sRGB doesn't have a pure gamma, but a combination of a short, gamma=1 segment tacked in front of a gamma=2.4 segment.

[smthopr]

I have come across strange results sometimes when converting from one profile to another which I believe are related to the Photoshop setting "preferences/performance/use graphics processor/advanced/use graphics processor to accelerate computation".

The solution is to uncheck this setting, which will result in slower performance, but without the "bug" if I recall.  I think you might have to relaunch Photoshop after changing this setting for it to take effect.

Try this and see if your results change and please report back here

[Doug Gray]

I have come across strange results sometimes when converting from one profile to another which I believe are related to the Photoshop setting "preferences/performance/use graphics processor/advanced/use graphics processor to accelerate computation".

The solution is to uncheck this setting, which will result in slower performance, but without the "bug" if I recall.  I think you might have to relaunch Photoshop after changing this setting for it to take effect.

Try this and see if your results change and please report back here

No question Photoshop uses different techniques with GPUs enabled for displaying but I haven't seen any profile conversion differences not associated with displays from using the graphics processors. The former drives me nuts dues to really bad conversions soft proofing with high bit monitors.

If you have an example of such a conversion difference I'd sure like to test it on my systems.

However, the OP's observations are fully explained by gamma differences.

[smthopr]

No question Photoshop uses different techniques with GPUs enabled for displaying but I haven't seen any profile conversion differences not associated with displays from using the graphics processors. The former drives me nuts dues to really bad conversions soft proofing with high bit monitors.

If you have an example of such a conversion difference I'd sure like to test it on my systems.

However, the OP's observations are fully explained by gamma differences.

Thanks for your reply. You may well be correct.

Can you please help me understand though... how a profile conversion does not adjust for gamma differences between color spaces?  I routinely print through .icc profiles from profoto RGB space and I think I would have noticed such a gamma shift in my prints.  I can't run an experiment now as my print has just died... but it will be replaced eventually!

[Doug Gray]

Thanks for your reply. You may well be correct.

Can you please help me understand though... how a profile conversion does not adjust for gamma differences between color spaces?  I routinely print through .icc profiles from profoto RGB space and I think I would have noticed such a gamma shift in my prints.  I can't run an experiment now as my print has just died... but it will be replaced eventually!

You are correct that printing adjust for different gammas in the image colorspace. That's actually the problem. The OP was printing patches from 1:255 but they started in different colorspaces:

one is Adobe-RGB  1-255
the other Prophoto-RGB 1-255

Whether you start with individual patches or make a gradient in Adobe RGB or ProPhoto, they will be gamma adjusted before printing. Or displaying for that matter.

Simple way to do this is to make two new images, one assigned to Adobe RGB and the other assigned to ProPhoto RGB. Then select all and fill wiith RGB (30,30,30).  View them side by side and the Adobe RGB one will be much darker. You can use the info tool to view the Lab values which will adjust any color space to Lab.

[kers]

Hello Doug ,

Thank you for explaining- i was puzzled.
I will be aware now that the RGBvalues represent very different LAB values between Adobe-RGB and Prophoto RGB.
The grayscale 1-255 of Prophoto has a more delicat gradual change in my opinion due to the lower Gamma, as i understand now.

Bruce, if your printer is not working- that does not matter; you can see the difference in photoshop.
I attached my grayscale i just made for checking my printer: it is in ProPhoto
Duplicate the file and change it into Adobe RGB ( not by conversion; just give it a new colorspace)
and see the difference. ( you can see the print difference via Softproofing)

[smthopr]

You are correct that printing adjust for different gammas in the image colorspace. That's actually the problem. The OP was printing patches from 1:255 but they started in different colorspaces:

Whether you start with individual patches or make a gradient in Adobe RGB or ProPhoto, they will be gamma adjusted before printing. Or displaying for that matter.

Simple way to do this is to make two new images, one assigned to Adobe RGB and the other assigned to ProPhoto RGB. Then select all and fill wiith RGB (30,30,30).  View them side by side and the Adobe RGB one will be much darker. You can use the info tool to view the Lab values which will adjust any color space to Lab.

Thanks. It makes sense now.  Two different images were printed as they were gradients created from two different color spaces with different gammas.