Working space and Colorvision Profiler Pro

[Greg]

Hi,
I'm using Profiler Pro V2.3 to profile my Epson 2200.

In the documentation, it says that during profile creation, the working
space in Photoshop should be the same
as the working space that will be used afterwards, while editing and
printing images. It also says that it is important
to always use this same working space, because otherwise there will be a
colour shift. This makes sense - I have read
elsewhere that profiles are (or can be) to an extent optimised for a
particular source gamut.

This is all fine, except that what I'm noticing is that the resulting
profile seems to behave identically, *regardless* of
the working space I use while building the profile! Has anyone else noticed
this? Should I tag all the Profiler Pro
ancilliary images (the images it opens up during the profiling, and which
are tagged by default with Adobe RGB)
with my new working space? Should I *convert* them all to my working space?

I am in touch with Colorvision support as well, but so far they don't even
seem to understand what my issue is.

Thanks,
Greg.

 

[Mike Russell]

Hi,
I'm using Profiler Pro V2.3 to profile my Epson 2200.

In the documentation, it says that during profile creation, the
working space in Photoshop should be the same
as the working space that will be used afterwards, while editing and
printing images. It also says that it is important
to always use this same working space, because otherwise there will
be a colour shift. This makes sense - I have read
elsewhere that profiles are (or can be) to an extent optimised for a
particular source gamut.

This is all fine, except that what I'm noticing is that the resulting
profile seems to behave identically, *regardless* of
the working space I use while building the profile! Has anyone else
noticed this? Should I tag all the Profiler Pro
ancilliary images (the images it opens up during the profiling, and
which are tagged by default with Adobe RGB)
with my new working space? Should I *convert* them all to my working
space?

I am in touch with Colorvision support as well, but so far they don't
even seem to understand what my issue is.

Photoshop is designed to recognize embedded profiles, and honor them for
printing and display purposes. The end goal of this is consistency, which
is what you are seeing.

There are ways to thwart this - for example by selecting the "Same as
source" option in the print dialog, and this may well be what the
Colorvision folks are refering to. For more detail, see:

<http://www.computer-darkroom.com/ps7_print/ps7_print_mac_2.htm>

 

[Greg]

Photoshop is designed to recognize embedded profiles, and honor them for
printing and display purposes. The end goal of this is consistency, which
is what you are seeing.

I understand this. When I said "identical results", I mean *exactly* - the
resulting pixel values are 100% identical.
This doesn't make sense to me at all. If Colorvision really mean what they
say, then when I build two different
profiles (using two different working spaces during the profile creation), I
would expect at least the perceptual
tables to produce slightly different colours, when I switch between the two
profiles. It appears to me that
it makes no difference whatsoever what my working space is set to during
profile creation - this is contrary
to the Profiler Pro documentation.

Greg.

 

[Johan W. Elzenga]

I understand this. When I said "identical results", I mean *exactly* - the
resulting pixel values are 100% identical. This doesn't make sense to me
at all. If Colorvision really mean what they say, then when I build two
different profiles (using two different working spaces during the profile
creation), I would expect at least the perceptual tables to produce
slightly different colours, when I switch between the two profiles. It
appears to me that it makes no difference whatsoever what my working space
is set to during profile creation - this is contrary to the Profiler Pro
documentation.

Perhaps the documentation is not correct in this respect. This is the
way I see it: First you print a color chart. It shouldn't really matter
in which color space you opened that chart, because in the end you will
print it in the *printer color space*. That's all the printer can do.
Next you measure the results by scanning that print. You build the
profile based on that, so I don't see any reason why the profile should
be different if the color space of the chart is different. As long as
you do not "Convert to profile" when you open the chart, the RGB values
you get (and so the RGB-values you sent to the printer) are identical,
no matter what color space you used.

 

[Greg]

Perhaps the documentation is not correct in this respect.

I've been kind of wondering whether this might be the case.

This is the
way I see it: First you print a color chart. It shouldn't really matter
in which color space you opened that chart, because in the end you will
print it in the *printer color space*.
Agreed/understood.

That's all the printer can do.
Next you measure the results by scanning that print. You build the
profile based on that, so I don't see any reason why the profile should
be different if the color space of the chart is different. As long as
you do not "Convert to profile" when you open the chart, the RGB values
you get (and so the RGB-values you sent to the printer) are identical,
no matter what color space you used.

I have read that it is possible to optimise the perceptual intent tables for
different assumed
*source* gamuts. For example, if the source gamut is large (e.g, Ekta
Space), the perceptual conversion
needs to scale the gamut down much more than if the source gamut is small
(e.g, sRGB). This is what I thought Profiler Pro might be trying to do -
optimise the perceptual tables for the user's working space. So, by some
means, I thought the Profiler Pro
plugin might have determined what the current working space is (during
profile creation), and taken that into account while
actually building the profile. The documentation is very clear in it's
directive to set the working space during profile creation to whatever
working space will be used afterwards. But as you say, it may just be
incorrect.

Greg.

 

[Greg]

As long as

Also, note that other images (two, I think) are opened by Profiler Pro,
*during* the profile creation. I don't
think these images are important - I think Profiler Pro just needs *any*
image open for certain
operations, simply because it's a Photoshop plugin. These images are tagged
with Adobe RGB,
and the documentation clearly says not to convert them when opening. (i.e,
to leave them in Adobe RGB).

Greg.

 

[Johan W. Elzenga]

Also, note that other images (two, I think) are opened by Profiler Pro,
*during* the profile creation. I don't think these images are important -
I think Profiler Pro just needs *any* image open for certain operations,
simply because it's a Photoshop plugin. These images are tagged with Adobe
RGB, and the documentation clearly says not to convert them when opening.
(i.e, to leave them in Adobe RGB).

I can understand that you shouldn't *convert* anything. That makes
sense. Let say that the color chart contains a patch with RGB values
200,200,200. You print this on your printer and next you scan this
print. The scan will be opened in your default color space, so that will
be AdobeRGB. Because the original chart was in AdobeRGB as well, the
software can make a direct comparision: The scanned patch should be
200,200,200 again, but it won't be. The difference is what is used to
make the profile. However, if you opened the chart and used "Convert to
profile" in order to use another color space, the original patch value
would not be 200,200,200 any longer. It will be converted to something
else, so you printed something else. Scan this into AdobeRGB, and you
obviously find a value that is not 200,200,200. Not because of the
printer, but because the value wasn't 200,200,200 to begin with. The
result is that your new profile will be off.

By the way, it seems that Profiler Pro works a little differently from
the one I got. I have PrintFIX and DoctorPRO. PrintFIX doesn't open any
other image during profile creation. DoctorPRO does open an image during
profile correction, but that seems indeed because something has to be
open (DoctorPRO runs an action, that can't run if there's no open
document).

 

[Greg]

By the way, it seems that Profiler Pro works a little differently from
the one I got. I have PrintFIX and DoctorPRO. PrintFIX doesn't open any
other image during profile creation.

FWIW, I've just tried it again, and it opens an image called
"Colorvision.tif" twice -
once when I open the measurements file, and once when I click the ok button
to
generate the profile. The image is the colorvision logo.

DoctorPRO does open an image during
profile correction, but that seems indeed because something has to be
open (DoctorPRO runs an action, that can't run if there's no open
document).

Indeed.

Greg.

 

[Johan W. Elzenga]

FWIW, I've just tried it again, and it opens an image called
"Colorvision.tif" twice - once when I open the measurements file, and once
when I click the ok button to generate the profile. The image is the
colorvision logo.

Could well be that this is for entertainment only. Personally, I can't
see why the Colorvision logo would be needed to create a profile!

 

[Pard]

I can understand that you shouldn't *convert* anything. That makes
sense. Let say that the color chart contains a patch with RGB values
200,200,200. You print this on your printer and next you scan this
print. The scan will be opened in your default color space, so that will
be AdobeRGB. Because the original chart was in AdobeRGB as well, the
software can make a direct comparision: The scanned patch should be
200,200,200 again, but it won't be. The difference is what is used to
make the profile. However, if you opened the chart and used "Convert to
profile" in order to use another color space, the original patch value
would not be 200,200,200 any longer. It will be converted to something
else, so you printed something else. Scan this into AdobeRGB, and you
obviously find a value that is not 200,200,200. Not because of the
printer, but because the value wasn't 200,200,200 to begin with. The
result is that your new profile will be off.

By the way, it seems that Profiler Pro works a little differently from
the one I got. I have PrintFIX and DoctorPRO. PrintFIX doesn't open any
other image during profile creation. DoctorPRO does open an image during
profile correction, but that seems indeed because something has to be
open (DoctorPRO runs an action, that can't run if there's no open
document).

But if your printer normally expects an sRGB (converted or otherwise
from Adobe RGB), doesn't it make sense to convert to sRGB anyhing sent
to the printer first? Of course you could set PS to work in sRGB, but
that seems counterproductive. Or maybe with Profiler, you just forget
all that sRGB business and use the profile that it creates?

Do you have a color reflection densitometer? Maybe this could be used
to verify that your Profiler program is doing what it's supposed to.
Get a known color reference chart and compare it to the printed
results from a scanned-in copy. Also there's no guarantee that your
scanner has perfect color, either--maybe this why Profiler has those
color tweaks available.

 

[Greg]

But if your printer normally expects an sRGB (converted or otherwise
from Adobe RGB), doesn't it make sense to convert to sRGB anyhing sent
to the printer first? Of course you could set PS to work in sRGB, but
that seems counterproductive. Or maybe with Profiler, you just forget
all that sRGB business and use the profile that it creates?

Do you have a color reflection densitometer? Maybe this could be used
to verify that your Profiler program is doing what it's supposed to.
Get a known color reference chart and compare it to the printed
results from a scanned-in copy. Also there's no guarantee that your
scanner has perfect color, either--maybe this why Profiler has those
color tweaks available.

I am using the raw mode of my Epson 2200, and it most certainly does not
expect
sRGB! The profiler is indeed working well - colours are accurate, greyscale
is neutral, and
I can print colours way outside sRGB. (e.g, Lab 41,-33,-36, confirmed by
measuring with a
colorimeter)

Greg.

 

[Johan W. Elzenga]

But if your printer normally expects an sRGB (converted or otherwise
from Adobe RGB), doesn't it make sense to convert to sRGB anyhing sent
to the printer first? Of course you could set PS to work in sRGB, but
that seems counterproductive. Or maybe with Profiler, you just forget
all that sRGB business and use the profile that it creates?

Of course you should forget about sRGB and use the printer profile you
just created. What is the use of making a printer profile if you don't
use it?

 

[Greg]

Could well be that this is for entertainment only. Personally, I can't
see why the Colorvision logo would be needed to create a profile!

No, I do not think it's for entertainment. I think it's simply that it needs
*any* image open,
just like Doctor Pro. However, it is possible that the colour space of this
image matters.

I've just dumped the perceptual tables (B2A0 tags) from the two different
profiles (one made with Adobe RGB set as my working
space, the other made with Ekta Space set as my working space), and they are
identical - the working space I used
made absolutely no difference to the resulting printer profile. Anyway, I'm
now talking to someone at Colorvision who seems much more technical than the
first person - I'll report back after we're done.

Greg.

 

[Pard]

I am using the raw mode of my Epson 2200, and it most certainly does not
expect
sRGB! The profiler is indeed working well - colours are accurate, greyscale
is neutral, and
I can print colours way outside sRGB. (e.g, Lab 41,-33,-36, confirmed by
measuring with a
colorimeter)

Greg.

That gives hope at least. Still it makes me wonder how best to use
sRGB, since most photolabs of the Frontier type expect sRGB images.
For example, is there any way to get bright reds out of these
machines? I send in a color chart and the red target is quite muted,
although the other colors and the grey scale look about right. Or is
this simply a characteristic of sRGB that must be lived with?

I can do better on my home printer (1280) than that, but not as
cheaply. You give values of 41, -33, -36, but these are RGB levels
you see on the screen--what are you getting for actual print color
values (in density) for the same colors, and are these outside the
sRGB gamut?

 

[Greg]

That gives hope at least. Still it makes me wonder how best to use
sRGB, since most photolabs of the Frontier type expect sRGB images.
For example, is there any way to get bright reds out of these
machines? I send in a color chart and the red target is quite muted,
although the other colors and the grey scale look about right. Or is
this simply a characteristic of sRGB that must be lived with?

Sorry I can't help with this.

I can do better on my home printer (1280) than that, but not as
cheaply. You give values of 41, -33, -36, but these are RGB levels
you see on the screen--what are you getting for actual print color
values (in density) for the same colors, and are these outside the
sRGB gamut?

The colour 41,-33,-36 are the L*a*b* coordinates of a colour which I have
measured with a colorimeter, from a print made with my Epson 2200. It is
way outside the colour gamut of sRGB, and most of the population's monitors,
in fact. It is a beautiful aquamarine colour.

Greg.

 

[Greg]

Just to bring a bit of closure to my original question - Colorvision say
that the working space *does* make
a difference, when using CMYK profiles. They haven't actually yet agreed
that there is no difference whatsoever
for RGB profiles, but they seem to agree that the documentation should
ideally be changed, to move the advice to use
whatever working space *during* the profiling, that is intended to be used
aftewards for actual
printing, to the CMYK profiling section of the document.

I can't use CMYK profiles so I'm not going to bother trying to verify this.

Greg.