Yves,
I ran an RGB test chart on my Epson XP-440 printer (can't go wrong with the
price of this printer, the ink is another story...) on a Canon Matte paper.
I first printed the chart from Photoshop.
When I opened the RGB TIFF, I assigned the "sRGB IEC61696-2.1" profile.
Then, I printed the chart the following way:
https://1drv.ms/u/s!AkD78CVR1NBqkodAnsrLlqoeSMMcSQ?e=xVldSE
Please note that "sRGB IEC..." and "EPSON sRGB" are the same "profiles".
Then I sent the same RGB TIFF from ACPU.
The result is here :
https://1drv.ms/u/s!AkD78CVR1NBqkodB0NMx1P2kjK-bxQ?e=UwkdRv
I also ran 'colverify.exe' on the two measurement files and got the following :
C:\Argyll_V2.1.1\bin>colverify -v ACPU.txt Photoshop.txt -k
argyll: Warning - Input file 'ACPU.txt' doesn't contain keyword DEVICE_CLASS
No of test patches = 288
File 0 Chose patch 270 as white, XYZ 0.828849 0.836080 0.801543
argyll: Warning - Input file 'Photoshop.txt' doesn't contain keyword
DEVICE_CLASS
File 1 Chose patch 252 as white, XYZ 0.829558 0.837229 0.801497
No of test patches in worst 10% are = 29
No of test patches in best 90% are = 259
Verify results:
Total errors: peak = 6.668890, avg = 0.845561
Worst 10% errors: peak = 6.668890, avg = 2.800466
Best 90% errors: peak = 1.566269, avg = 0.626672
avg err X 0.002770, Y 0.002738, Z 0.003257
avg err L* 0.371701, a* 0.448715, b* 0.441771
I agree there are some rather large differences between the two outputs but if
you look at this screen capture :
https://1drv.ms/u/s!AkD78CVR1NBqkodCOtFYmU8Ii8EUSg?e=oirnVn
You will see that the differences are shown with "red outlines", these
represent DeltaE differences > 3.0. To me, the "difference" between those two
prints are attributable to a combination of the Epson printer and the Canon
paper, and not to a difference between printing from Photoshop "direct", using
a null conversion, and printing from ACPU. But, as someone famous on the
ColorSync List, "Your mileage may vary".
BTW#1, I tried to print from GIMP but I got an error on the "Print Plug-in"?
Maybe some other time...
BTW#2, Photoshop is 'smart' enough to detect when the same profile is used for
both the 'Input' and the 'Output', as shown here:
https://1drv.ms/u/s!AkD78CVR1NBqkoc_bbaXLLGqFqo3dg?e=QDgC8X
It will flag the error to the user and if the user clicks the 'Import' button,
it opens up a new browser window to download ACPU (!)... It's possible to
ignore the error but, in my case, since I did not want to have the error show
up at the moment of printing, I assigned "sRGB IEC..." to the document and
selected "EPSON sRGB" in the Print dialog, this way, Photoshop was "happy" that
the two profiles were "'different".
"sRGB IEC..." and "EPSON sRGB" are the same internally. So this is in effect a
null transform.
Now, having written all that, I'm surprised (and disappointed?) that my Epson
"superlative" XP-440 printer performed so poorly, with color management turned
off? I shall try to redo the experiment using a less porous substrate...
/ Roger