Color management
Ramblings about Color Management
Sunday, February 1, 2009
To many photographers Color Management is more like black magic. It seems confusing and they don’t really understand the theory and science behind it. One recent internet posted called it a “mess” , falsely believing no one really understood it because there is so many contradicting statements floating around.
Invariably in every photography forum I visit there are threads about color management, full of all kinds of advice and theories. While there is certainly some good information, there is plenty which is not very good, and some of it is just down right bad. The “debate” rages on, with plenty of theories and claims out there, many of which are based on personal success with a unique workflow. It seems when someone who is struggling with color management finds something which works for them, they think it must be the “right way” and are quite happy to declare that in various forums and in the blogisphere. This turns to criticism of what may be correct principles ... after all, if they are right then everyone else is wrong. Eventually these discussions turn into rather heated debates without much substance, and sometimes even evolve into what I’ve dubbed Color Management Urban Legends - or CMUL’s for short. And thus we now understand the force behind the “mess” mentioned by that internet poster - a lot of misinformation by well meaning people who think they know what they are talking about, but perhaps don’t.
The problem with CMUL’s is they all seem logical. For example, my CMUL #1 is - If your output is only to an sRGB device such as the web or a photolab, then you should use sRGB as your working space. That certainly sounds logical, but in fact it is over simplistic and may result in lower quality of your final images. In addition there is no advantage at all in using sRGB in this case, but there certainly is potential drawbacks. There are several of these CMUL’s floating around, but before tackling them we need to have a pretty good grasp of color management. Otherwise the surface logic of these CMUL’s will cause us to miss the point, as so many others have.
I will admit there are a few rare circumstances where workflow can be optimized by taking direct control of a process that is very repetitive, extremely consistent, and frequent. However, even though it may work in these situations it should never be advocated for typical use. If you really want to skip color management, I would recommend you at least understand it and make sure what your are doing really does have some advantage.
So the bulk of this article is a general discussion of color management. Let me begin by stating for the record I am not an expert in color management. I am not a color scientist. My only claim to expertise in this subject is from a working photographer that has struggled for many years to grasp it, and finally believe I understand the underlying principles of applying good color management and use it successfully on a daily basis. I have been involved in photography and color printing for well over 30 years, doing both high end custom portrait photography and lab printing, as well as high volume photography and lab production. I certainly don’t want to be labeled as one of those contributing to the unnecessary confusion and proliferation of these CMUL’s, but I do believe my perspective will be helpful to anyone struggling to make sense of it all.
The overall goal of a color managed workflow is fundamentally very simple ... to allow devices of varying capabilities to display results that match - or at least appear to match. A lofty goal no doubt, and one that on the surface appears to be a pipe dream. If we as humans saw everything EXACTLY as it exists, this goal would be impossible. The reason it actually works, and works pretty well, is we don’t see that way. Human vision is very complex ... way over my head. But I know what we see is an interpretation of the information our eyes send to our brains, not a factual representation of the information.
A couple of quick and easy examples everyone is familiar with. If a single 60 watt lightbulb is supplying light in a room, and another 60 watt bulb is added, the amount of light is doubled, but we do not perceive the room to be twice as bright. If our vision were strictly linear, we could never handle the extreme difference between bright sunlight and low lit situations. Another example is our perception of white (and other colors for that matter). As long as the color temperature of the light isn’t too extreme (sunlight vs tungsten) we perceive white nearly the same. Only if there are conflicting sources of light can we detect the difference.
Because of this it is possible that two devices with quite different output characteristics can actually appear similar to our eyes. A good example is CMYK printing. We know it is possible to print images that are stunning and pleasing, in a book for example, despite the fact that CMYK printing on an offset press is a pretty low gamut device. This principle applies to nearly all output devices ... with a decent output device and a good color management system we can modify the information so the device appears much closer to what we expect. Despite the gamut differences of a monitor and a high end inkjet printer, images from each can look very similar. To our eyes, it’s the relationship of things like color and brightness that counts ... not their actual values. If we can characterize a device so we understand what it does when we send certain colors to it, we can use that information to modify the colors sent to the device and simulate those relationships, achieving a reasonable match - even if one device has much lower capabilities than another. From most photographers perspective, the devices we seek to match are our monitor and our printer. Indeed that is the holy grail of color management for photographers ... to accurately predict from the monitor what the print will look like.
To accomplish this requires setting up our color management correctly, some discipline in it’s use, and some trust. When things don’t go as expected, it is tempting to experiment, and it seems those that experiment and find something that works think they’ve solved a mystery and start preaching their solution. I don’t have a problem with anyone doing what they want, I just don’t think they should recommend their hack of color management as a general solution, and perpetuate all of the Color Management Urban Legends that are proliferating throughout the internet. This approach is typically successful because they simply impose an artificial limit somewhere in the color management process which doesn’t really fix the problem ... it just covers it up. They would be better off if they really found the true cause of their problem and solved it correctly, instead of using color management incorrectly to impose this artificial limit which just hides the problem.
There are several key things required to have a good color managed workflow, especially for those striving for the highest image quality.
1. A good monitor. While an Eizo or high end NEC designed for color critical work is nice, it certainly isn’t necessary. Many LCD monitors will do just fine. Having a higher gamut monitor may make things look a little better, but remember the very premise of color management is one device can be characterized and the data sent to that device in a way so that it will appear visually very similar to a higher gamut device despite it’s limitations. There is one caveat you should be aware of when choosing a monitor ... if you want a good monitor to print match you have to be able to control how bright the monitor is. Some brands and models are just too bright no matter what you do, making it very challenging to get the correct density of the final prints.
2.A good viewing station. If you want prints to match the screen, then you must view the prints in a consistent manner, using a light source that is full spectrum. I use color corrected florescent lamps (not ideal, but they work). I’ve heard that Solex lights work very well, and I may try them. I have a GTI print viewer next to my main workstation where I view my test 8x10 prints. I feel my screen to print match is very good for good papers that use photo black inks. It’s helpful if you can adjust the brightness of your viewing station without altering the color of the light. This makes matching densities between screen and print easier, since you can control both screen brightness and print brightness. But it’s not required. In this article from The Online Photographer is a suggestion for a pretty inexpensive idea for a viewing light - The Online Photographer: The Viewing Station. In the article comments there are also some suggestions.
3.A good monitor profile using the correct luminance. There really isn’t a way to do this without a good device and good software. There are several out there, but I can only speak for one ... the xRite i1 Pro and xRite ColorMatch software. The most common mistake here is the brightness of the monitor. If we want to reach our goal of screen to print match, we have to make sure that “white” from both devices appears the same. Almost all monitors are way to bright for this, and the key is setting our brightness so it correctly matches our viewing conditions. The setting involved here is called luminance (a fancy word for brightness, and the recommended starting point is around 120 cd/m2. On many monitors this is only about one-half of the maximum, and some monitors can’t even be dimmed this low. We can use our profiling tool to hit a target luminance, but ideally we should open a new document in Photoshop, place a piece of unprinted paper (the kind we normally print on) in our viewing station and adjust the monitor (and or viewing station if it is adjustable) until the two appear similar in brightness. It is important you do this before you make your monitor’s profile.
4.Good output profiles. Pretty hard to predict results if the color management system really doesn’t have a good idea of what to do with the data it is sending to the printer. With current high end inkjet printers canned profiles often suffice, since these devices have achieved much better consistency. With output to labs using photo printers it gets a little more challenging. A good lab will be applying an output profile to your image and can provide you with that profile so you can use Photoshop’s soft proof functionality. Many photographers make the mistake of thinking that the output profile is sRGB because that’s what the lab asks you to send. Somewhere along the line those colors are going to be remapped .... either using a profile or by some proprietary system. When I print an inkjet for someone, all I need is their file is tagged to the working space they were using. I don’t ask them to convert their file to my printer space. Many photolabs make the mistake of forcing you to move your file to sRGB and then they will move it to the printer space. Often this is because of the printers proprietary software is requiring sRGB so they have no choice. It would be better if they could move from the workspace to the output space in a single step, and I’m seeing this trend in better labs now, allowing AdobeRGB files. In reality they shouldn’t care what space the file is in - as long as they know which space they can correctly convert the file into the printers space.
5.Correct application of color management principles in your workflow. This means understanding what a working space is, how color management is applied, and how to use soft proofing to even better predict what your print will look like when using some types of output (especially matte surface papers).
It sounds a little daunting, but in reality it’s sort of a connect the dots thing. If you do everything correctly, it will just work. If you understand a little about why it works, you will quickly see the flawed logic of those proponents of various CMUL’s.
Well, enough for now. In part 2 I plan to build on this and describe the Ultimate Color Managed Workflow.
Aspen Ridge
Hasselblad H1 with PhaseOne p45 back, HC 150mm. 0.8 seconds at f/5.6, ISO 100
This image is made from 6 vertical captures, stitched with Photoshop in a single image.