Really confused about color space

When opening in Photoshop an image and then later assigning to it a different profile makes the picture look absolutly horrible. I noticed this when opening an image using the AdobeRGB profile and later assigning to it the ProPhoto profile. The colors were so saturated!
The right procedure is that when you open the picture you should select to convert the colors to the other space or else later on to convert (not to assign) to the other color space. The Photoshop 7 help says:

When using the Assign Profile command, you may see a shift in color appearance as color numbers are mapped directly to the new profile space. Convert Profile, however, shifts color numbers before mapping them to the new profile space, in an effort to preserve the original color appearances.

Hope this helps.

You have to assign a profile if the image isn't tagged with a profile, which probably doesn't happen too frequently if you are working on your own images.

Doesn't it take RGB values and change them to different values? Isn't the file a list of which changes to make?

My impression is that calibrating a monitor causes it to display standard colors correctly as closely as possible. Profiling a monitor creates a file that coverts colors being sent to the monitor such the displayed color matches the standard color.

A file's color space is a definition of the boundaries of displayable colors. Assigning a new color space changes the meaning of the RGB values, while converting to a new color space changes those values directly.

What I haven't figured out yet is why one printer profile that I paid real money for causes many greens and browns to block up horribly. I would have thought a profile would not change the way the printer lays down ink, it would just change the way the color values in the file are converted on the way to the printer. And that profile could also be used by Photoshop "in reverse" to convert the standard display to the printer's gamut.

I have been using Colorblind ProveIt with a Sequel Chroma 4 colorimeter to calibrate and profile my monitor, and was having fits with it (it's a new monitor). I finally gave up on it and profiled my monitor visually, and now it's bang-on, except around the boundaries of the gamut. Grrr!

When I have all that working, I keep all my working space in Adobe1998. If I'm going to display on the web, I use the web display tool in Photoshop to create a JPEG with no color management information in it. To make that work, I have to save my corrected file, make a new file targeted for web display, turn on the "Proof Colors|Windows RGB" in Photoshop, and then correct the now washed out colors. Otherwise, my wide-gamut Adobe numbers mean something not as wide in sRGB, with the result that the sRGB device displays it as less colorful. I have a feeling that the reverse is what is happening to the OP. I haven't tried to print color-matched stuff at labs, though.

One of these days I'm just gonna hire one of you guys to come to my house and set up my color workflow. I just have a master's degree in engineering and do system's engineering for a living. It's obviously not enough to understand color management!

Rick "who makes it work but never seemingly on purpose" Denney

One thing that would make the profiles much more interesting and understandable would be if they were ascii text files instead of binary files. Even better would be if they were using XML. From everything that I can discern from hacking them, they simply contain transfer function curves implemented as an array of discrete points.

Rick,

A profile does not convert colors, it just describes the color performance of a device or an arbitrary working space.
It is the job of the CMM(Color Management Module, Color Matching Module, etc...) to do the actual converting by the rules of rendering intents.

Analogy.
Take the entire English speaking world.
For the most part you can communicate with everyone in this group and understand what is meant. This represents an overview of modern color management. For the most part everything can communicate together and get the jist of what color is being communicated.

But there are variances of English that can screw things up if translated literally, mainly slang. For example "pissed" in America means angry, in England it means drunk. This is analogus to profiles of 2 different monitors American English slang would be a profile while Queens English slang would be another profile.
The fact that you can get the jist of what is being said by all these English speakers means there is a common translation point, the English language of course, and the knowledge of the subtle variances of the two, i.e a slang dictionary.
The English language and its dictionaries are the CMM of the linguistic world. Translating the subtleties of one version of english ,Queens English(one profile), to another form of english, American English(another profile), so that none of the meaning of the original message is left out.

This is an over simplified analogy, but it hits home.
Spoken language is no different from color language when you break it down.

I get that.

But then why does that expensive printer profile I bought result in splotchy ink patches in green and brown? The results are unusable, even though skin tones are more like what I see on my screen. If I only shot in high key, it might be okay.

If the idea is for the color engine to use the profile to convert the printer to a standard color presentation, so that 128,128,128 is the same on the printer as it is in the standard, then why does it take so much fooling around? Why can't the manufacturer just do that? Yes, I realize papers are different. But not THAT different.

I have calibrated and profiled my monitor so that it displays the standard colors, or so I'm led to believe by my monitor profiling sofware.

Your language analogy is a good one, in that it captures both the basis of communication (common language) and the concept of gamut (vocabulary). Choosing a color space with a smaller gamut is like using a smaller vocabulary. It works fine for common colors but might not be able to convey fringe ideas as effectively. Translating from one color space to another is a matter of making simple words complex or the reverse. But trying to interpret one color space when the file is built in a different color space is like trying to understand a thesis written to make use of a 10,000-word vocabulary with only 5,000 words at one's disposal. Much gets lost. Okay, enough of that.

I think I understand what is supposed to be going on. But when I bought my profile, I thought I would provide a way to proof my printer's output (i.e., display what my printer will print on my calibrated and profiled monitor), not change the way the printer lays down ink. I would think it's the manufacturer's responsibility to design the printer so that its output is close to standard color as possible.

Rick "who went back to the Epson profiles despite that they aren't quite right" Denney

Rick,

Quick question.
What printer and what software did you get a profile produced for?

Got some answers for you but need to know this first...

Epson 1270, using Conetech profiles. Windows XP with Photoshop CS. My profiling software is Colorblind Proveit! which came with a Sequel Imaging Chroma 4.

Actually, I'm now getting reasonable (not perfect) consistency between Photoshop's Proof Colors display and the printer. The "proof colors" setup uses the Epson Stylus Photo 1270 profile (which was supplied by Epson), with "preserve color numbers" checked as well as "paper white". As I understand it, that bypasses the "intent" engine, and that was a critical step for me in making things work.

My color settings include the Adobe 1998 color space and I preserve embedded profiles (my camera and scanner also put output in the Adobe 1998 color space). The engine is Adobe ACE and the intent is perceptual.

The printer driver is set up to use the EE089_1 profile, which is one of the two or three that I got from Epson.

I had calibrated and profiled my monitor using the Chroma 4, but my flat panel just refused to provide a reasonable gamma display on visual confirmation of the results. I therefore experimented with the "visual pathways" in the Colorblind software instead of using the Chroma 4. The results were vastly, vastly better. I use an Ott viewing light and have little trouble with light pollution in my work area. The white seems quite consistent from the Ott light to the monitor, and I adjusted a white display to match the viewing light using the monitor's controls before doing the profiling.

So, it's working, but only by breaking the rules. The expensive printer profile didn't work at all, and neither did the expensive monitor color sensor. And by "working" I mean that it's...close. But making subtle and final color adjustments requires test prints.

Rick "who realizes he should upgrade to a newer color profiler and an R2400 printer" Denney

Rick,

In my eyes you are doing quite well!
And I also wish that the printer/RIP people would try to make things easier by actually following a universal convention of color mamagement, ICC specs and stick with them. No proprietary hocus-pocus.

Here is the interesting part. Be cautious with this info, since I have not personally used the model of printer you are using.

The custom profie you had made probably/may will not work correctly.
Why? Well the print driver/RIP for most desktop printers require "secret-sauce" in the profile to work correctly.
Example. The RIP software that came with our Roland 8-color printer, ColorChoice, can use custom profiles but if they are not made with the special software for creating profiles for ColorChoice the lc, lm O & G will not fire, creating wierd results. The sofware required to create profiles that will work with ColorChoice adds all sorts of code that only ColorChoice will recognise and translate.
Why do printer/RIP companies do this? Why can't they be straight up ICC compliant?
They probably want to drive a portion of the population insane!
Blows me away.

Also, the media/ink/resolution combination that you make a profile for makes a HUGE difference. It is valid for only that combination, and nothing else. Even matte vs. semi gloss media will make a huge difference. You can use a profile for a combination that it was not initially intended for, but the results will not be what you expected - worse or better-in-some-cases! This scenario mainly applies to canned profiles, depending upon your ambient conditions, humidity - temp., etc...

Down in San Diego today and need to get ready to watch the Chargers B-B-Q the Chiefs!

UncleDoug wrote:
In my eyes you are doing quite well!

You response calms my nerves considerably. With this stuff, I'm not qualified to be a theoretician so I have to depend on experience alone. And that puts me squarely in the aphorism of Poor Richard: "Experience keeps a dear school, but fools will learn in no other."

But my description of what I did didn't offend those whose opinion I respect, then I won't worry that I've just used a Band-Aid to cover some other critical mistake I've made.

I didn't have the profile made for me specifically. It was made for my printer and media combination (actually I bought different profiles for a range of Epson media). And I bought the monitor profiling sofware and sensor from the same folks. If I attempt it again using the officially correct method, I'll go with someone else. For now though, it's working about as good as I think I'm going to get.

Rick "appreciative for the advice" Denney

Back to Amy's original post,........ I've been plagued by the same problem for months. I shoot RAW Adobe RGB, and process in CS3 Adobe RGB. They look great on the monitor, but when printed, or viewed in any other application they look amber and muddy.

Last Friday, I purchased a very good book (don't know if the forum allows me to post the title, but I'd like to shout praises for the author!). The difference is the intent of the images. sRGB was written for web viewing, and Adobe RGB is for printing. Now, I can relate to the problem, but I still don't have a cure. I am using a lab that requires .jpg files in sRGB.

If Adobe RGB, or even ProPhoto RGB is the color manager for print, why would the labs (which many pros seem to use) require an sRGB image?

Rebecca

Have a read in the link from my sig....

Hi Rebecca. Even given all the help I got here, I was never able to solve my problem. It simply was not a common problem I guess. I even tried completely reinstalling Photoshop with no luck. Since that time, I bought a new computer and at the same upgraded to Photoshop CS3 and WALLA the problem was gone. Now I can happily convert from SRGB to Adobe RGB without the orange tint showing up in my prints. I never figured out what was wrong. I hope you have better luck!

Hi Rebecca, you're most welcome to mention the book - we're always eager to know of good resources.

Your lab is no different to most others - they all require sRGB. Even if you find one that accepts other colour spaces, they are simply converting down to sRGB or an sRGB-esque profile.

sRGB is your friend - embrace it.

adjohnson wrote in post #7060646
Hi Rebecca. Even given all the help I got here, I was never able to solve my problem. It simply was not a common problem I guess. I even tried completely reinstalling Photoshop with no luck. Since that time, I bought a new computer and at the same upgraded to Photoshop CS3 and WALLA the problem was gone. Now I can happily convert from SRGB to Adobe RGB without the orange tint showing up in my prints. I never figured out what was wrong. I hope you have better luck!

In this case, judging from the images here:

Robert_Lay wrote in post #989694
So far as I know, that's the way it's supposed to work. Remember, within Photoshop, I never see this difference!

I'd say that it was a case of comparing a color managed program to a non color managed program.

Both images (AdobeRGB and sRGB) look identical in FF3 with color management enabled. They obviously will look different in a non color managed application.

If the printer prints sRGB images different from AdobeRGB images, he's probably ignoring the color profile. Bad.

If your images change appearance in Photoshop when converting from AdobeRGB to sRGB, something is wrong. They shouldn't. (And from what I read in the OP, they don't)


Have a look in the link from my sig.