I have another thread started here about setting up for costco. I do not want to divert it with this question.

I have CS3, calibrated monitor, generally work with 16bit adobe rgb images from a D300 and print, using the appropriate profile, on an R1800.

I and others have noted that such prints are generally darker than the screen would indicate they should be.

I note that if I soft proof to the epson profile, I get a much better visual representation of what I will be getting on the print.

I thought the job of the profile was to translate my file so the print would match what I saw on my calibrated monitor. To my way of seeing this, either the calibrated monitor is grossly over rated or the profiles are borderline worthless, yet many of you swear both are necessary and vital.

Can you account for this paradox? Am I doing something radically wrong?

There's no paradox. You simply can't accurately soft proof without good calibration, good profiles and good viewing conditions. Compromise on any of them and you'll suffer less than stellar results. I sort of remember your other thread, but I don't recall if you actually got your calibration issues sorted or not.

All I can tell you is that with great calibration and custom profile, my screen to print matches are scary. I've had dozens of so-called printing industry professionals tell me it could never work, and then have to eat there words when they see it in person.

There are so many variables including your ambient lighting that in order to help you out we're going to need all the details...


Peter

Color management for printing starts with the output devices and works backwards to allow the display to simulate the limits of the output. The goal for soft proofing is to have the screen accurately simulate printer output, in effect forcing more saturated colors of the monitor to look as "bad" as the print will.

The management process starts by profiling the printer using a target with as many as 900 different color patches and then comparing how the printer ink/paper reproduced them with how they would look if reproduced optimally "per the numbers". The gamut of a printing device is dictated by the physical characteristics of the paper and pigments. The red, green, blue phosphor / LCD pixel on your monitor will usually be capable of reproducing much more saturated reds, blues and purples than any paper / pigment combination because the inks are actually transparent and magenta and cyan pigments are cross contaminated. The resulting printer profile, when used for soft proofing, provides a "roadmap" telling the monitor how to desaturate its color and lower its contrast on screen to simulate how the print output will look.

In the days before ICC profiles what we'd do is take a file containing the test chart RGB values and send it to the printer, then manually adjust the printer color and contrast controls until the image on the screen was as close as possible match to the print sitting in a 5000K standard viewing conditions. The result would be an appearance on screen that was far less optimal than the screen could make it, but one which provided some clue of how the colors would look when printed. You'd need one monitor for normal viewing and another for "soft proofing".

What ICC color management does is dynamically alter the output characteristics of the monitor to simulate any gamut. The Catch-22 is that what it displays is limited by its gamut. If you edit on a consumer grade LCD monitor or laptop your "window" on the color values which actually exist in the file are rather limited to a gamut similar to sRGB or even smaller. So while you have mapped the camera color values to the rather large AdobeRGB working space, what you are seeing during normal editing are any more saturated colors than your monitor can reproduce remapped to fit the gamut of the monitor in a way that makes the overall contrast and saturation of the image look good PERCEPTUALLY.

The reason a large working space is needed is because RGB and CYMK colorspaces, when mapped to the three coordinate Lab mathematical model of human color perception are much different shapes..

Chuck,

I really can't thank you enough for all of your contributions. This is yet another excellent post that has taught me much that I thought I already knew - but obviously didn't.

It is really appreciated.

Mike.

And yet, Chuck's overly complicated response still does little or nothing to help Papageno with his problem. Why do Papageno's prints look darker than his screen?

The most likely reasons are that he has poor calibration on a LCD screen that is too bright in terms of candelas per meter squared. Unless your calibration software lets you set a numerical value for cd/m2 that is somewhere in the 110-120 cd/m2 region, the chances are very good that your screen is set way too bright. You are then adjusting your images on a too bright screen to make them look good to your eye, and of course, they're going to look dark.

But in reality, it's not the overall screen brightness that matters, it's the ratio of screen brightness to ambient room illumination that matters. With CRTs we (and I still do) kept our offices darkened because the CRTs could only hit 95 or 90 cd/m2 without prematurely burning themselves out. If you raise your ambient room illumination, that will compensate for the brighter LCD screens by forcing your iris's to close down somewhat. It's still a matter of trial and error to come up with the right combo that works for you. There are also better and worse hardware calibration systems out there. They are NOT all created equal.

The second factor is that you absolutely need good quality profiles, both monitor and output for the soft proofing to work well. Until we hear the specifics of Papageno's setup, it's impossible to say exactly where the problems lie. If you're using a canned profile, particularly one grabbed online for some Costco somewhere, I would be doubly suspect. You have to know for a fact that the profile you are using is in fact one made not only for the printer you are actually using, but that the printer is in the same state of calibration and operating with the same driver options as when it was profiled.

Seriously, it ain't all that complicated. Good hardware calibration. Good profiles. Proper room illumination. Appropriate print viewing conditions.

First of all, many thanks to cgardner for taking the time post a wonderful answer that clearly took a lot of time and effort to prepare. Kudos!

With regard to Peter Figen's remarks, the monitor is an NEC 2470 calibrated with Spectravision. cd/m2 is about 95. Costco provides location specific profiles from dry creek; this particular store provides a "better" profile. On the other hand, the machines are profiled every six months, in which time, much can happen.

The 1800 profiles I use are both custom and canned; what is striking to me is that all match very well. I can make the same print on glossy galerie, epson super duper premium luster and enhanced matte and the prints are remarkably close. This consistency leads me to think the profiles are good.

Room illumination is by 5000 degree high cri (over 90) fluorescents which are shielded from the screen. The screen is about a stop lighter than the ambient light.

I thought the job of the profile was to translate my file so the print would match what I saw on my calibrated monitor. To my way of seeing this, either the calibrated monitor is grossly over rated or the profiles are borderline worthless, yet many of you swear both are necessary and vital.

Psychologically, I know I like the richness I see on screen, which of course can not be fully captured on paper, That said, after eyeballing the print to my idea of visual perfection, I almost always need to lighten it 1/4 to 1/3 of a stop if I use the exposure slider, or a similar amount using other methods. Soft proofing on the 1800 is pretty much accurate no matter what paper I use.

It may not be that complicated, so perhaps I can't see the forest for the trees....

papageno wrote:
Psychologically, I know I like the richness I see on screen, which of course can not be fully captured on paper, That said, after eyeballing the print to my idea of visual perfection, I almost always need to lighten it 1/4 to 1/3 of a stop if I use the exposure slider, or a similar amount using other methods. Soft proofing on the 1800 is pretty much accurate no matter what paper I use.


It appears you have your workspace and workflow under good control. Since soft proofing works fine for you with your in-house Epson 1800 workflow the culprit is obviously the Costco profile which is not matching the results the printer is actually producing.

Why is that happening? The problem reading the tea leaves here is your description of what you are seeing. You use generalities like "richness" and "light and dark" and are not telling us what specific parts of the tonal scale are being rendered differently. If print making exposure is the root cause of the difference between what the profile predicts and actual results the exposure variance from profile would manifest itself over the entire tonal range of the image: the highlights would be darker than they appear in the soft proof and the shadows would be just as dark but carry less detail in the 3/4 tones. But a shift in the midtone values can also cause a similar perceptual difference in tone and "richness".

1/4 to 1/3 stop is well within the range of variation you'll encounter with photographic products. To get a really accurate profile you'd need to make one from each new batch of paper. There should be some way for the operators at Costco to test each new batch of paper and re-calibrate the baseline exposure setting for the machine, but that may not be happening.

The other possibility is that the operators are not turning off all the corrections and you are getting some form of double color management. There are apparently different ways to "turn off corrections" on the Noritsu printers and the Dry Creek tutorial mentions the need to be very specific about instructions to the operator.

Costco printers don't recognize color space tags. Thus if you were to send a file you edited in AdobeRGB to Costco for printing the color management engine in the printer would conform it to the native color space of the printer much i the same way your Epson would if you set the driver to "Printer Manages Color". In theory that should produce prints with the maximum saturation the pigments in the paper are capable of. By applying the Costco profile before printing you change the actual file values sent to the printer and telling the operator to turn off corrections is equivalent to "Photoshop Manages Color" and that only works well when the profile exactly matches the printing conditions.

To make head or tails out of what is happening I would suggest printing some images of a standard target like the Macbeth color checker. If you shot it in the camera and then soft proof and print it would give you a better idea of where on the tonal scale the changes are occurring and if some colors are affected more than others.

Chuck

Chuck

Thanks for the excellent explanation and example to go with it.

So one of the questions is, how does one determine the "quality" of the monitor calibration? Is there an objective way? In other words, I am NOT interested in proof 5 photos and compare type of answers which will be very subjective based on the photos I print.

-- V

Vivek wrote:
Chuck

Thanks for the excellent explanation and example to go with it.

So one of the questions is, how does one determine the "quality" of the monitor calibration? Is there an objective way? In other words, I am NOT interested in proof 5 photos and compare type of answers which will be very subjective based on the photos I print.

-- V

In practical terms the baseline for your color management workflow is your camera, the source of the images. If you go outside on a clear sunny day and shoot a Macbeth ColorChecker with the camera set on Daylight WB or custom WB set off a gray card and the exposure set so the white patch on the target is about 1/3 stop below clipping and open it on your monitor you want it to look similar to how it did by eye. If it does then you might reasonably say the quality of your monitor calibration is OK. Another empirical way to judge the calibration of your monitor vs others is to download test images from various sources.

In photographic reproduction two separate color management workflows are needed. One workflow needs to be optimized to make the color look as good as possible on the screen. That's the default workflow cameras and monitors are optimized for. The second workflow needs to be optimized for printing to recognize and compensate for the inherent physical / technical limitations of the process. What the ICC profile based color management does, in theory (and practice if well managed) is to make the process of managing the two different workflows transparent to the end user. But the end users can't be completely clueless because managing both workflows effectively starts with the selection of the working space for editing. It needs to be large enough to hold the gamut of the printer and the gamut of the monitor which are different shapes. Its like finding single box "one size fits all" box that will hold either an apple or a potato which have might have similar volumes but different shapes.

The idea of soft proofing to predict print results has been the "Holy Grail" of color management since ColorSync and ICC profiling was introduced in the early 1990s as a way to eliminate the very costly process of making pre-press proofs. My work background is in offset printing and starting in the 1970s. In the commercial realm a proof OK from the client is like a contract. The printer in producing the proof is saying it represents what the final production run will look like. The customer in signing off on the proof acknowledges they realize the final printed result might not match the range of color and saturation in the screen image (or transparency in days past).

I've seen proofing evolve from single color progressives printed on flat-fed presses only a generation removed from Gutenberg, through various forms of pre-press proofing such as Cromalin and Matchprint, into the digital realm first with dye sub printers like the 3M Rainbow and now ink jets. There are many variables in the printing process which affect outcome and the most reliable predictor of final results is a proof made with the same final film which is used to expose the plates for the production run. Proofing would usually occur in two stages: loose proofs of individual color subjects, then a final "form" proof after the pages are laid-out in 8 page segments or forms for each side of the press sheet. Soft proofing can work quite effectively in a closely controlled in-house workflow such as in a printing company, but it requires expensive equipment for creating accurate profiles and knowledgeable staff to run it. We use it for our in-house printing operation at the Dept. of State, but only internally. We still have customers sign-off on hard copy proofs because we can't control or predict their viewing conditions.

Once you grasp that the goal of color management in the context of the commercial publishing / printing environment it originated was as a communication tool to lower expectations regarding the final output its easier to understand the idea behind soft proofing in Photoshop. The proof is rarely the absolute best results the proofing method can produce, just a simulation of the more limited capabilities of the CYMK production equipment used for the final output. Soft proofing a photo before printing -- seeing the image through the "filter" of the printer profile --tempers your expectations of what the print will look like. Ideally, with an accurate press profile the soft proof image on the screen will match the print, but neither will match an image optimized for viewing on a web page.

The needs of most photographers are much simpler than in commercial publishing and printing -- pleasing realistic looking color -- but the same technical limitations apply. The main difference is that the whole host of technical problems inherent in the process of converting an RGB file into a CYMK printed page aren't on the radar screen of most photographers. They send an RGB file to the printer and it automatically comes out as a CYMK version of the same image without any work or technical skill required on their part. As with most things which become automatic push-button operations they stay blissfully ignorant about how difficult it is technically to convert a RGB image into a CYMK one and to varying degrees develop unrealistic expectations.

CYMK printing works because the inks are transparent. The full spectrum illuminant passes through the thin ink film and the inks pass the some wavelengths and block others. The wavelengths which are passed through the ink layers then bounce off the paper, back through the ink and then hit the eye of the viewer. The need keep the inks transparent limits the amount and types of dyes and pigments which can be used and the maximum density which can be printed while still retaining shadow detail. So to produce a printed image in which all the colors appear natural and the steps of a gray scale appears neutral requires a good bit of complicated technology under the hood of the printer.

The limiting factor in the print workflow are the colors 100% of the individual inks and overlapping combinations can produce. To visualize the limits of your printer and understand what soft proofing is doing try this simple exercise. In Photoshop create a file which has solid patches of 100% Red, Green, Blue, Cyan, Magenta, and Yellow then print it. Then hold it up next to the screen. That will give you a clue how different the outer limits of the RGB and CYMK gamuts are. Next add a Hue /Saturation adjustment layer and try to visually match the image on the screen to how the print sitting next to it looks. What you'll be creating manually via that adjustment layer will have the same net effect as applying the printer profile to the screen image in soft proofing mode.

That exercise is more for edification than any practical use. In the process of making the adjustments in saturation of the various colors to match screen to print results you'll begin to get a clue how the RGB and CYMK gamuts differ. The colors in which you need to reduce saturation to get a match are the ones your screen can display but your printer will never be able to print. But if you take the print into the other room and look at it. It will look OK. Wait until dark and look at in under the fluorescent lights in the kitchen and under the tunsten lighting over your bathroom sink. It will look OK there too. The magic which makes color reproduction work is the fact human vision is so highly adaptive that absent any direct side-by-side comparison our eyes "see" what our brains expect them to see. Absent anything in your field of vision with a more saturated red than is printed on the page your brain will tell you the red patch on the print its a really nice rich red. Just don't compare it to a can of Coke or a Ferrari.

So in the final analysis if it looks good, it is usually is good and there's a rather large range of tolerance which viewers will as "normal". When you need to worry is if you post a photo which looks fine on your monitor and you get comments from others that it has a color cast.

Chuck

The soft proof makes allowances for the paper and ink combination that you will be printing with. You will never get as bright an image from reflected light on paper as you do from the transmitted light of the screen. Perhaps if you print on film and use a backlight, you will have a similar effect.

Barry

Will run a few from the fourth with comparison in mind and report.

This is one of the best threads I've seen here; thanks to you all!

I am resolved to get some improvement.

After two days of fumblefingered experimenting....

The current settings for calibration are 5500 degrees kelvin, gamma of 1.8 and white of 65 cd/m2. Not perfect, but getting close.
Soft proofing for Costco lustre at Mount Prospect shows one significant change; a considerable loss of saturation. Adding saturation at a setting about 17 seems to bring it back.

If I'm getting workable results through a combination of unlikely errors, I'd like to know what they are.

I should add that Delta e is .25 lower than I've ever had it......

Why gamma 1.8? 2.2 is industry standard.

Maybe.....but I've been trying as many variables as possible to help get this monitor so that what I see is what I get.

It doesn't matter how much horsepower the engine has if it wont start.......

Current setting is more accurate--for me---than anything I have tried.....

At this point, Papageno, it would be best to start again with a detailed list of exactly where you're at and what you are specifically doing. Trying different things willly-nilly, as you appear to be doing is only going to get you to a place where you have no idea where you are or how you got there. It might even be better to just give you a very specific to do list and have you follow it to the letter, see that it does work, and then work backwards to understand why.

Peter

This centers on my NEC 2470 monitor and the calibration via spectravision. The calibration has consistently resulted in a monitor that is too bright.

Put another way, when the image is right on visually, on the monitor, it prints darker--on my Epson 1800 (using what I believe are solid, consistent profiles); at Costco using their specific profiles and or rare occasions various other printers.

Since my work environment is 5000 degree high CRI fluorescents, it seems logical to work with 5000 degrees........

I frankly do not understand the function of the gamma setting, merely assumed lower gamma meant darker.....

If you would be so kind as to send specific starting points, I;'d be happy to try them; strange as it may sound, I normally appreciate the idea of doing things correctly.

One other note: I appreciate everything that has been written here about the assorted gamuts/color spaces available. That said, I think that part of the discussion is to some degree a red herring. Correct color isn't the issue here; lights and darks are.

What would really be helpful is a set of specific settings for both cs# and spectraview, including whether I should set or reset anything on the 2470 in the manual settings.....

Email or here would really be useful.

Also, someplace that lists direct settings.....

Thanks again!