Discussion:
ICC profile connection space
(too old to reply)
Dale
2012-10-27 07:37:53 UTC
Permalink
the current ICC profile connection space is an ideal print, or maybe an
ideal subtractive system http://www.color.org

for transparent display, translucent display, motion picture display and
other additive systems like CRTs LCDs etc. this means you have to create
information to render the ideal print to these spaces

wouldn't it be better too have an ideal additive system as the profile
connection space and compress the tone and color the data for prints, as
opposed to creating information for additive systems

probably would lead to more precise capture rendering too
--
Dale
Thomas Richter
2012-10-27 13:24:41 UTC
Permalink
Post by Dale
the current ICC profile connection space is an ideal print, or maybe an
ideal subtractive system http://www.color.org
for transparent display, translucent display, motion picture display and
other additive systems like CRTs LCDs etc. this means you have to create
information to render the ideal print to these spaces
wouldn't it be better too have an ideal additive system as the profile
connection space and compress the tone and color the data for prints, as
opposed to creating information for additive systems
probably would lead to more precise capture rendering too
I beg your pardon, but one of the choices for the profile connection
space is XYZ which is a perfectly linear, additive system.

Greetings,
Thomas
Dale
2012-10-27 14:27:27 UTC
Permalink
Post by Thomas Richter
Post by Dale
the current ICC profile connection space is an ideal print, or maybe an
ideal subtractive system http://www.color.org
for transparent display, translucent display, motion picture display and
other additive systems like CRTs LCDs etc. this means you have to create
information to render the ideal print to these spaces
wouldn't it be better too have an ideal additive system as the profile
connection space and compress the tone and color the data for prints, as
opposed to creating information for additive systems
probably would lead to more precise capture rendering too
I beg your pardon, but one of the choices for the profile connection
space is XYZ which is a perfectly linear, additive system.
Greetings,
Thomas
isn't this the XYZ relative to an ideal print? I'm pretty sure it is.
There is also the difference between color matching and appearance
matching. Appearance matching can be image dependent so you would have
to use some kind of scene balance algorithm, SBA , which doesn't quite
work for professional images and is best left to the image
editor/artist. SBA can prove good enough for most consumer images.
--
Dale
Thomas Richter
2012-10-29 08:48:13 UTC
Permalink
Post by Dale
isn't this the XYZ relative to an ideal print?
No. XYZ is a perfectly additive space based on physical intensities and
spectral curves based on work of the CIE.
Post by Dale
I'm pretty sure it is.
You are confused.
Post by Dale
There is also the difference between color matching and appearance
matching.
Which is irrelevant for the question of the profile connection space. It
is the matter of how an ICC color management module implements the
conversion between source and target space.
Post by Dale
Appearance matching can be image dependent so you would have
to use some kind of scene balance algorithm, SBA , which doesn't quite
work for professional images and is best left to the image
editor/artist. SBA can prove good enough for most consumer images.
And your point is?
Dale
2012-10-27 14:56:38 UTC
Permalink
Post by Thomas Richter
Post by Dale
the current ICC profile connection space is an ideal print, or maybe an
ideal subtractive system http://www.color.org
for transparent display, translucent display, motion picture display and
other additive systems like CRTs LCDs etc. this means you have to create
information to render the ideal print to these spaces
wouldn't it be better too have an ideal additive system as the profile
connection space and compress the tone and color the data for prints, as
opposed to creating information for additive systems
probably would lead to more precise capture rendering too
I beg your pardon, but one of the choices for the profile connection
space is XYZ which is a perfectly linear, additive system.
Greetings,
Thomas
not exactly, you can't have an absolute profile connection space and
have appearance matching, you need a reference medium for the profile
connection space, which there is

D50 would be a better choice for an ideal additive reference system, it
could be used for some systems looking beyond the visual spectrum also,
whereas XYZ is limited to the visual spectrum
--
Dale
Dale
2012-10-27 15:29:59 UTC
Permalink
Post by Dale
beyond the visual spectrum
the infrared

and the ultraviolence

http://en.wikipedia.org/wiki/Clockwork_orange

"Does God want good, or the choice of goodness, is the man who chooses
the bad, in some way better than the man who has the good forced upon him."
--
Dale
Dale
2012-10-27 15:52:32 UTC
Permalink
Post by Dale
D50 would be a better choice
strike that, sorry, needs XYZ and a reference additive display device,
like a television, motion picture projector, computer monitor, etc.
--
Dale
Dale
2012-10-27 19:20:45 UTC
Permalink
Post by Dale
Post by Dale
D50 would be a better choice
strike that, sorry, needs XYZ and a reference additive display device,
like a television, motion picture projector, computer monitor, etc.
or maybe D50, XYZ, and an ideal transparency or set of filters
--
Dale
Thomas Richter
2012-10-29 08:52:12 UTC
Permalink
Post by Dale
not exactly, you can't have an absolute profile connection space and
have appearance matching, you need a reference medium for the profile
connection space, which there is
D50 would be a better choice for an ideal additive reference system, it
could be used for some systems looking beyond the visual spectrum also,
whereas XYZ is limited to the visual spectrum
D50 is not a reference system. It is a white point. XYZ is an absolute
color system - you only need a white point if you need a white point
adaption, i.e. if you want to map the scene white point to the target
(display device, printer etc) white point. Not said that you shouldn't
do that, but that's a different task and part of what a CMM might or
might not implement, for example by means of a Bradford transformation.

Of course XYZ is "limited" to the visual spectrum, that's the whole
point of XYZ - what is visible by humans. If you need color matching for
your canary bird, well, XYZ and ICC profiles are not for you. But I
believe that then the market for such technology is pretty limited. (-:
Dale
2012-10-30 11:50:34 UTC
Permalink
Post by Thomas Richter
Post by Dale
not exactly, you can't have an absolute profile connection space and
have appearance matching, you need a reference medium for the profile
connection space, which there is
D50 would be a better choice for an ideal additive reference system, it
could be used for some systems looking beyond the visual spectrum also,
whereas XYZ is limited to the visual spectrum
D50 is not a reference system. It is a white point. XYZ is an absolute
color system - you only need a white point if you need a white point
adaption, i.e. if you want to map the scene white point to the target
(display device, printer etc) white point. Not said that you shouldn't
do that, but that's a different task and part of what a CMM might or
might not implement, for example by means of a Bradford transformation.
Of course XYZ is "limited" to the visual spectrum, that's the whole
point of XYZ - what is visible by humans. If you need color matching for
your canary bird, well, XYZ and ICC profiles are not for you. But I
I will restate, it is the intent and working space I have a problem with
and in so the CMM

the perceptual and saturation and relative colorimetric intents are
based on an ideal print, absolute colorimetric is not,

absolute colorimetric intent might have a use case for converting
between the same devices, that is all I see

my problem with working spaces is most applications (I use gimp) don't
allow CIELAB, CIELUV or CIEXYZ working spaces

and gimp doesn't allow you to choose the connection space, I think this
is specified in the profiles not the CMM
--
Dale
Thomas Richter
2012-10-31 07:40:33 UTC
Permalink
Post by Dale
I will restate, it is the intent and working space I have a problem with
and in so the CMM.
the perceptual and saturation and relative colorimetric intents are
based on an ideal print, absolute colorimetric is not,
Not really. Perceptual and saturation are not based on "print", but
"reproduction by an output device". Whether that is a monitor or a
printer is irrelevant for the purpose of a CMM. Neither would I say
"ideal" because it is the job of the profile to compensate for the
non-idealness of the device. If the device and the profile do not fit to
each other, then this is not the failure of the CMM, but the vendor by
providing you a profile that is wrong.
Post by Dale
absolute colorimetric intent might have a use case for converting
between the same devices, that is all I see
No, it is just a matter of what your intent is. Absolute is just one
possible intent, i.e. get exactly the same colors. That means, of course
due to adaption of the human eye, that the colors will look different
under different illumination, even for devices that create their colors
themselves (such as monitors), but it is of course more a problem for
devices using a multiplicative color reproduction (such as printers).
Post by Dale
my problem with working spaces is most applications (I use gimp) don't
allow CIELAB, CIELUV or CIEXYZ working spaces
Huh? First of all, the choice of the PCS does, ideally, not change the
rendering. The PCS is just the coordinate system within which the
profiles are specified, and the PCS is used as a common fixpoint between
input and output device, nothing more. Thus, up to numerical errors,
whether the PCS is XYZ or CIELab shouldn't make a difference, just that
some profiles and rendering intents are more easily expressed with
CIELab than XYZ.

Second, if gimp is your problem, why not report to the gimp authors?
Nobody here can help you with the deficiencies of gimp (only 8 bit per
sample, no ICC support). You can either a) report to the gimp team, or
b) supply patches to fix the problem, or c) use another program, but
your approach d) complain about it here is not very efficient and won't
help at all.
Post by Dale
and gimp doesn't allow you to choose the connection space, I think this
is specified in the profiles not the CMM
Why should gimp do that? And what would be the purpose of that? Yes, the
rendering intent is in the profiles, but it is up to the CMM to support
other ones and reflect the choices of the user. Once again, the PCS does
not encode a rendering intent, nor should it make any visible difference
- it is only a coordinate system.

Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
Dale
2012-11-01 08:52:07 UTC
Permalink
Post by Thomas Richter
Why should gimp do that? And what would be the purpose of that? Yes, the
rendering intent is in the profiles, but it is up to the CMM to support
other ones and reflect the choices of the user. Once again, the PCS does
not encode a rendering intent, nor should it make any visible difference
- it is only a coordinate system.
okay the PCS isn't the problem alone, but the PRM, print reference
medium is, and the PRM is what you get in the PCS unless you choose
absolute colorimetric intent

I think ProPhotoRGB PCS was chosen to try to deal with higher gamut
spaces, but RGB isn't the best choice, should have chosen a CIE
coordinate system, like XYZ
--
Dale
Dale
2012-11-01 08:53:18 UTC
Permalink
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
--
Dale
Thomas Richter
2012-11-01 10:01:38 UTC
Permalink
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
I don't get you. You get what you pay for. Either gimp does what you
need to do, then use it and be quiet. Or it doesn't, but then look for
alternative solutions. *Here* nobody can help you with gimp. Wrong place
for requesting improvements.
Dale
2012-11-01 11:07:05 UTC
Permalink
Post by Thomas Richter
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
I don't get you. You get what you pay for. Either gimp does what you
need to do, then use it and be quiet. Or it doesn't, but then look for
alternative solutions. *Here* nobody can help you with gimp. Wrong place
for requesting improvements.
maybe gimp people are listening
--
Dale
Dale
2012-11-02 13:06:50 UTC
Permalink
Post by Thomas Richter
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
I don't get you. You get what you pay for. Either gimp does what you
need to do, then use it and be quiet. Or it doesn't, but then look for
alternative solutions. *Here* nobody can help you with gimp. Wrong place
for requesting improvements.
what if linux supercedes mac and windows, then what?

photoshop doesn't have a linux version, besides gimp is free

gimp has an official IRC chatroom, who is to say they aren't lurking here?

http://www.gimp.org/

http://www.gnu.org/

http://www.ubuntu.com/
--
Dale
nospam
2012-11-02 14:20:28 UTC
Permalink
Post by Dale
what if linux supercedes mac and windows, then what?
that's so not going to happen. anyone who thinks it's even remotely
possible is living in fantasy land.
Post by Dale
photoshop doesn't have a linux version,
one reason of many why it won't ever happen.
Post by Dale
besides gimp is free
so what?
Post by Dale
gimp has an official IRC chatroom, who is to say they aren't lurking here?
that won't add the missing features.
nospam
2012-11-01 16:32:15 UTC
Permalink
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
gimp isn't even worth free.
Dale
2012-11-02 13:01:58 UTC
Permalink
Post by nospam
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
gimp isn't even worth free.
for free its more than a heckuva program
--
Dale
nospam
2012-11-02 14:20:25 UTC
Permalink
Post by Dale
Post by nospam
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
gimp isn't even worth free.
for free its more than a heckuva program
not if it doesn't do what you need.
Dale
2012-11-04 14:36:59 UTC
Permalink
Post by nospam
Post by Dale
Post by nospam
Post by Dale
Post by Thomas Richter
Second, gimp doesn't allow you to use a fully calibrated color workflow,
it only operates in the 8bpp "color space" your monitor seems to have.
But if that's your problem, I suggest to buy more professional software
that offers support ICC profiles.
gimp is free, and even better the upgrades are free
gimp isn't even worth free.
for free its more than a heckuva program
not if it doesn't do what you need.
I really have no issues with gimp, I just forgot about using ProPhotoRGB
as the working space for higher gamut images and output

sorry, forget the whole thread
--
Dale
Dale
2012-11-01 08:54:53 UTC
Permalink
Post by Thomas Richter
Post by Dale
I will restate, it is the intent and working space I have a problem with
and in so the CMM.
the perceptual and saturation and relative colorimetric intents are
based on an ideal print, absolute colorimetric is not,
Not really. Perceptual and saturation are not based on "print", but
"reproduction by an output device". Whether that is a monitor or a
printer is irrelevant for the purpose of a CMM. Neither would I say
"ideal" because it is the job of the profile to compensate for the
non-idealness of the device. If the device and the profile do not fit to
each other, then this is not the failure of the CMM, but the vendor by
providing you a profile that is wrong.
Post by Dale
absolute colorimetric intent might have a use case for converting
between the same devices, that is all I see
No, it is just a matter of what your intent is. Absolute is just one
possible intent, i.e. get exactly the same colors. That means, of course
due to adaption of the human eye, that the colors will look different
under different illumination, even for devices that create their colors
themselves (such as monitors), but it is of course more a problem for
devices using a multiplicative color reproduction (such as printers).
Post by Dale
my problem with working spaces is most applications (I use gimp) don't
allow CIELAB, CIELUV or CIEXYZ working spaces
Huh? First of all, the choice of the PCS does, ideally, not change the
rendering. The PCS is just the coordinate system within which the
profiles are specified, and the PCS is used as a common fixpoint between
input and output device, nothing more. Thus, up to numerical errors,
whether the PCS is XYZ or CIELab shouldn't make a difference, just that
some profiles and rendering intents are more easily expressed with
CIELab than XYZ.
Second, if gimp is your problem, why not report to the gimp authors?
Nobody here can help you with the deficiencies of gimp (only 8 bit per
sample, no ICC support). You can either a) report to the gimp team, or
b) supply patches to fix the problem, or c) use another program, but
your approach d) complain about it here is not very efficient and won't
help at all.
Post by Dale
and gimp doesn't allow you to choose the connection space, I think this
is specified in the profiles not the CMM
Why should gimp do that?
because you want control over that aspect of the workflow, for instance
if an input profile asks for one thing and an output profile asks for
another it is left up to the CMM to choose the PCS
--
Dale
Thomas Richter
2012-11-01 09:58:38 UTC
Permalink
Post by Dale
Post by Thomas Richter
Post by Dale
and gimp doesn't allow you to choose the connection space, I think this
is specified in the profiles not the CMM
Why should gimp do that?
because you want control over that aspect of the workflow, for instance
if an input profile asks for one thing and an output profile asks for
another it is left up to the CMM to choose the PCS
You still don't understand the role of the PCS then. Once again,
changing the PCS should not change the rendered result. It is just a
matter of the coordinate system, but not the matter of the resulting
transformation.
Dale
2012-11-02 05:49:37 UTC
Permalink
Post by Thomas Richter
Post by Dale
Post by Thomas Richter
Post by Dale
and gimp doesn't allow you to choose the connection space, I think this
is specified in the profiles not the CMM
Why should gimp do that?
because you want control over that aspect of the workflow, for instance
if an input profile asks for one thing and an output profile asks for
another it is left up to the CMM to choose the PCS
You still don't understand the role of the PCS then. Once again,
changing the PCS should not change the rendered result. It is just a
matter of the coordinate system, but not the matter of the resulting
transformation.
look here http://www.color.org/v4_prmg.xalter
--
Dale
Thomas Richter
2012-11-02 07:57:30 UTC
Permalink
Post by Dale
look here http://www.color.org/v4_prmg.xalter
What has this to do with the PCS? It is just a refinement how to
implement the perceptual rendering intent.
Dale
2012-11-02 13:01:11 UTC
Permalink
Post by Thomas Richter
Post by Dale
look here http://www.color.org/v4_prmg.xalter
What has this to do with the PCS? It is just a refinement how to
implement the perceptual rendering intent.
the PCS is dependent on the intent
--
Dale
Thomas Richter
2012-11-02 18:29:01 UTC
Permalink
Post by Dale
Post by Thomas Richter
Post by Dale
look here http://www.color.org/v4_prmg.xalter
What has this to do with the PCS? It is just a refinement how to
implement the perceptual rendering intent.
the PCS is dependent on the intent
Where did you get this from??? No, it is not!
"Perceptual Reference Medium" != "Profile Connection Space".
Dale
2012-11-04 14:36:22 UTC
Permalink
Post by Thomas Richter
Post by Dale
Post by Thomas Richter
Post by Dale
look here http://www.color.org/v4_prmg.xalter
What has this to do with the PCS? It is just a refinement how to
implement the perceptual rendering intent.
the PCS is dependent on the intent
Where did you get this from??? No, it is not!
"Perceptual Reference Medium" != "Profile Connection Space".
I didn't say equal, I said dependent, all intents except the absolute
colorimetric choice are dependent on the reference medium, and I believe
this means XYZ, LAB, sRGB, ProPhotoRGB, etc.

I really have no issues with gimp, I just forgot about using ProPhotoRGB
as the working space for higher gamut images and output

sorry, forget the whole thread
--
Dale
Dale
2012-10-28 01:23:06 UTC
Permalink
Post by Thomas Richter
Post by Dale
the current ICC profile connection space is an ideal print, or maybe an
ideal subtractive system http://www.color.org
for transparent display, translucent display, motion picture display and
other additive systems like CRTs LCDs etc. this means you have to create
information to render the ideal print to these spaces
wouldn't it be better too have an ideal additive system as the profile
connection space and compress the tone and color the data for prints, as
opposed to creating information for additive systems
probably would lead to more precise capture rendering too
I beg your pardon, but one of the choices for the profile connection
space is XYZ which is a perfectly linear, additive system.
Greetings,
Thomas
not a good encoding or calculation space, need E*

probably added to the mix for people who want to avoid the right
calculations
--
Dale
Loading...