Photoshop Manages Color - Pointless?

[Doug Gray]

Well sure I don't understand everything it does. Not completely clueless so I think we can hardly say it's a waste of money. I don't believe in the concept of FULLY knowing everything to start using it and learning from it while slowly understanding its technical aspects. That's like saying one cannot walk without fully understanding how to walk. It kind of just develops through trial and error like anything else I'd expect.

It's a very good tool and there is nothing better to increase understanding of color printing than a tool that you can use to check your process. It's like reading about how a mill or lathe works. Nothing like actually having them when reading about how to use them.

[r010159]

Hey Everyone,

So I'm sort of confused. If the printer...already prints the way it does...and a printer profile simply just takes a differential measurement between what your printer prints and ICC accepted standards of target palette colors...then what is the point of using Photoshop Manages Colors and a Printer Profile?

Does that somehow make colors more accurate, and if so  - WHY?

The printer already prints whatever colorspace a certain way..so what on earth can the ICC Printer Profile do differently when coupled with Photoshop Manages Colors to get prints more accurate?

If you like the results from printer managed color, then do not bother wth a color managed workflow. I am sure if you make enough prints of an image, you will figure out how to make the result look attractive.

Bob

[Steve Upton]

It's really quite simple and not worth worrying about too much.
When a printer profile is made a spectrophotometer measures the colours the printer actually prints, the profiling software compares them to what colour they should be, then creates the profile to try to make the printer deliver as many colours as it can, as accurately as it can.

Unfortunately this is really not how it works.

I'm all for simplification and analogies but if they create an idea in people's minds that is incorrect then I don't think they're a good idea.

In 2003 I wrote an entry in our "Color Myths" series about this very topic.

The process of building a printer profile samples the colors produced when device values (RGB, CMYK, whatever) are sent to a printer and measured. Then, in the future, when a color value (Lab) is desired on the printer, the printer profile outputs the device value that will produce that color (or something close that agrees with other colors in the process).

It may sound like splitting hairs but I assure you, it's different. And for those who choose to study the process further, they won't come to a confusing point and have to unlearn their initial ideas behind the process.

To return to the OP's post:

If the printer...already prints the way it does...and a printer profile simply just takes a differential measurement between what your printer prints and ICC accepted standards of target palette colors...then what is the point of using Photoshop Manages Colors and a Printer Profile?

Does that somehow make colors more accurate, and if so  - WHY?

So, as I mentioned above, the printer profile is not about a differential measurement. It's about rendering the color you want in the best way possible.

If you let the driver perform the conversion then it uses its own internal profile. If you are happy with the output of the printer in it's default behavior, then you're good. But be assured, a profile is being used in either case.

But if you find your shadows are plugged, your highlights are blown out, your reds are tomato, your blues are purple or any of a large number of color rendering problems, then a custom profile could very well do a better job getting you the color you want. And letting Photoshop perform the conversion is more likely to get the results you want.

[Rhossydd]

It may sound like splitting hairs

It is ABSOLUTELY splitting hairs.
Bringing things like 'LAB values' into trying to explain how the absolute basics of custom printer profiling works to someone without any idea is not at all helpful.

[Steve Upton]

It is ABSOLUTELY splitting hairs.
Bringing things like 'LAB values' into trying to explain how the absolute basics of custom printer profiling works to someone without any idea is not at all helpful.

Ok, remove that word from my description. Then I think it stands pretty well as a lay-person description of the process.

Also I'd prefer to leave it to the OP to decide if it's useful.

[Simon Garrett]

If you like the results from printer managed color, then do not bother wth a color managed workflow. I am sure if you make enough prints of an image, you will figure out how to make the result look attractive.

Bob

With the Epson printers I've used (R2000, R1800) you can have colour-manage workflows with either the program or the printer (driver) implementing colour management, provided you enable that function in the printer driver.  The results are near-identical. 

I use colour management in the program (Photoshop or Lightroom) as it's more convenient, but both work. 

[EricV]

In theory, to obtain a color managed workflow, we would like each device to calibrate/profile itself.  So the monitor should do whatever is necessary to display RGB values from a file with absolutely accurate color, and the printer should do whatever is necessary to print RGB values from a file with absolutely accurate color.  Of course absolute accuracy is not always possible, since gamuts do not completely overlap, but that is the ideal.  An application like Photoshop is then responsible only for producing files with correct RGB values in a properly tagged color space. 

The same result can be obtained by letting an external application change the RGB values that it sends to the printer, while telling the printer it is no longer responsible for color management.  For this to work, the external application must know the characteristics of the printer.  This breaks the paradigm of device-independence and seems rather backward.  Why should Lightroom know how your Epson printer renders RGB values into ink colors better than the printer itself?  Nevertheless, this has historically been the case, and most of us have gotten used to working this way, because it gave better and more reliable results.  If printers are finally becoming capable of handling their own color management, it would be good to be able to go back to that model. 

Maybe a contrived example would help.  Suppose your printer prints RGB=(128,128,128) with a blue tinge when color management is turned off.  (Assume the color space defines this color to be neutral gray.)  You create a printer profile to compensate for this.  The profile says something to the effect that in order to print (128,128,128) accurately, the printer must be told to print (128,128,112).  That profile can be loaded into the printer, with color management turned on, so that when it receives (128,128,128) it attempts to print (128,128,112), which comes out as neutral gray.  Or the profile can be applied by the application, so that it translates (128,128,128) into (128,128,112) before sending the image to the printer, with color management turned off.  The end result should of course be the same.

[GWGill]

It is ABSOLUTELY splitting hairs.
Bringing things like 'LAB values' into trying to explain how the absolute basics of custom printer profiling works to someone without any idea is not at all helpful.

Introducing the concept of a Device Independent intermediate colorspace is essential for curing the common misconception shown by the OP that color profiling is about "correcting the devices color".

[GWGill]

In theory, to obtain a color managed workflow, we would like each device to calibrate/profile itself.  So the monitor should do whatever is necessary to display RGB values from a file with absolutely accurate color, and the printer should do whatever is necessary to print RGB values from a file with absolutely accurate color.

ICC style color management doesn't work like this, because it isn't that narrow or fixed in function.
You are assuming some sort of definition of how an RGB value "should" be displayed or printed. No such thing exists in relation to a single (general purpose) input or output device.  They behave how they behave - profiling captures that and deals with making them do something useful at a higher (system) level. Rather than repeat myself, you should read this.

[Tim Lookingbill]

Of course we must consider the non-uniform nature of Lab space as a intermediate color describer which happens to have a history of issues with blues turning purple. So nothing's perfect in this coding color box. Just FYI. Not arguing the matter.

[GWGill]

Of course we must consider the non-uniform nature of Lab space as a intermediate color describer which happens to have a history of issues with blues turning purple.

Now that really is a quibble - any slight non-uniformity in L*a*b* is insignificant compared to it's huge increase in perceptual uniformity compared to XYZ, and of absolutely no consequence as a space to represent color value in, where it has the property of great efficiency. It only has any consequence if used as a gamut mapping or clipping space.

[hugowolf]

And remember, in Ps, at the user end at least, Lab values are integer. which is not a good thing.

Brian A

[GWGill]

And remember, in Ps, at the user end at least, Lab values are integer. which is not a good thing.

L*a*b* values (like all other colorspace values) are whatever precision you choose them to be. In 8 bit the a* & b* are integer and the L* has 1 binary bit. If you choose 16 bit, they are not integer.

[Tim Lookingbill]

So are all ICC profile transforms calculated with 16 bit precision?

[Doug Gray]

So are all ICC profile transforms calculated with 16 bit precision?

As a practical matter, yes.  ICC profiles do allow 8 bit LUTs for printers but I've never run across one. Standard profiles are calculated using 16 bits but the larger source of error is the LUT interpolation.

[Lundberg02]

I tried to tell the nut case who runs Photography Life that. No use.

[GWGill]

So are all ICC profile transforms calculated with 16 bit precision?

The precision used by the profile has nothing directly to do with the precision used in transforms, although it does limit the overall precision of course.

For instance, ArgyllCMS's icclib has always used floating point for transforms, and so has lcms from V2.

[Simon J.A. Simpson]

Unfortunately this is really not how it works.

I'm all for simplification and analogies but if they create an idea in people's minds that is incorrect then I don't think they're a good idea.

In 2003 I wrote an entry in our "Color Myths" series about this very topic.

The process of building a printer profile samples the colors produced when device values (RGB, CMYK, whatever) are sent to a printer and measured. Then, in the future, when a color value (Lab) is desired on the printer, the printer profile outputs the device value that will produce that color (or something close that agrees with other colors in the process).

It may sound like splitting hairs but I assure you, it's different. And for those who choose to study the process further, they won't come to a confusing point and have to unlearn their initial ideas behind the process.

To return to the OP's post:

So, as I mentioned above, the printer profile is not about a differential measurement. It's about rendering the color you want in the best way possible.

If you let the driver perform the conversion then it uses its own internal profile. If you are happy with the output of the printer in it's default behavior, then you're good. But be assured, a profile is being used in either case.

But if you find your shadows are plugged, your highlights are blown out, your reds are tomato, your blues are purple or any of a large number of color rendering problems, then a custom profile could very well do a better job getting you the color you want. And letting Photoshop perform the conversion is more likely to get the results you want.

So, for the sake clarity and the removal of doubt, do I understand correctly that you are saying the you are using the word “calibrate” to mean ‘measure and adjust to an external standard’ (e.g. colour temperature) whereas (and put simply) you are saying that a device profile (e.g. that of a printer) describes the behaviour of the device which allows colour management software (e.g. ‘Adobe ACE’ colour engine) to produce as accurate appearing as possible a range and relationships between colours that the device can produce/encode; and (for the sake of completeness) in order to facilitate this there are, in addition, a number of ways of interpreting this “accuracy” which are the Rendering Intents (e.g. “Perceptual”, “Absolute”, etc..) ?

Without getting too technical have I, in essence, understood correctly ?  I am happy to be corrected where I may have got it wrong but colour management pedants please desist from being picky !   

[Simon Garrett]

So, for the sake clarity and the removal of doubt, do I understand correctly that you are saying the you are using the word “calibrate” to mean ‘measure and adjust to an external standard’ (e.g. colour temperature) whereas (and put simply) you are saying that a device profile (e.g. that of a printer) describes the behaviour of the device which allows colour management software (e.g. ‘Adobe ACE’ colour engine) to produce as accurate appearing as possible a range and relationships between colours that the device can produce/encode; and (for the sake of completeness) in order to facilitate this there are, in addition, a number of ways of interpreting this “accuracy” which are the Rendering Intents (e.g. “Perceptual”, “Absolute”, etc..) ?

Without getting too technical have I, in essence, understood correctly ?  I am happy to be corrected where I may have got it wrong but colour management pedants please desist from being picky !   

Calibration doesn't normally apply to printers.  Printers are profiled, which means measuring the colour space of the printer (usually the printer/paper combination).   

Calibration normally applies more to monitors, where calibration and profiling are two processes, normally done at the same time by the software that comes with the colorimeter or photometer.  Here's what happens with monitors:

• The software calibrates (adjusts) the monitor to a known white point and tone response curve (TRC, or gamma).  This will be with a combination of hardware adjustments to the monitor and a Look Up Table (LUT) that's loaded into the driver, and is used by the driver to adjust the TRC and white point.  Normally the colour space is fixed, but some wide-gamut monitors can emulate a narrower gamut, and for those monitors the colour space itself can be calibrated (e.g. to sRGB).
• After calibration, the monitor is profiled (measured): the colour space and calibrated TRC and white point are measured, and that measurement is put in the profile. (The LUT created by calibration is normally also saved in the profile, but it isn't technically profile information.)

Printers are just profiled, they are not normally calibrated first. 

Colour-managed programs, when they output to a monitor or printer, take the profile and map colours from the image colour space (sRGB, Adobe RGB or whatever) into the monitor's or printer's colour space, as defined in the profile. 

[GWGill]

So, for the sake clarity and the removal of doubt, do I understand correctly that you are saying ....

See What's the difference between Calibration and Characterization ?