Color Management Myths 16-20
by Steve Upton
Myth #16: Two 5000K bulbs will match
First let's talk about 5000K, D50, D65, etc, how they are specified and what they describe.
5000K is a color temperature. It's a color of white that is based on heating a "black body" object to 5000 Kelvin. When you get something this hot it glows a yellowish-white. This color is expressed as a colorimetric number, such as XYZ, Yxy or Lab and is then called a correlated color temperature. The lights that produce these colors of white are not actually burning at that temperature, they just make the same color of light as a black body that is - that's the correlation. D50, D65, etc are daylight illuminant standards based on spectral curves.
Remember, spectral curves are the most detailed light information we can get. When we add human observer influence we get colorimetric numbers. An important thing to realize is that many different spectral curves can result in the same colorimetric number. Another way of putting this is to say that fluorescent, incandescent and sunlight can all have significantly different spectral output but still produce the same color of white for a human observer; the same colorimetric number.
This "match" is for an even white card that has no brighteners. As each of these light sources have a different spectral make-up they will affect papers containing brighteners and printed colors differently. The net effect of this is that two bulbs that measure as 5000K can affect paper and printed color differently and that one print can look different under two 5000K bulbs.
There is another measurement that reflects how closely a light source matches the D50 daylight spectral curve. CRI, the Color Rendering Index is expressed as a value out of 100. 100 is a perfect match to a reference light source that is within 100K temperature (D50 in the case of standard viewing environments). Typical cool white fluorescent bulbs have a CRI of about 62. The CIE states that the color differences between sources that differ by 3 CRI units is not perceptible. You should aim for CRIs that are as high as possible (over 95) when shopping for lighting and viewing products - and the booth can affect it as well.
I suppose the summary point of this myth is that 5000K is not a guarantee that you have a good light source or that it will match another 5000K light source effectively.
Myth #17: D50 bulbs are available
D50, as mentioned above, is a standard daylight spectral curve.
This spectral information is defined very specifically using more than 30 curve points. It has a single colorimetric number and, unlike 5000K, it cannot be created by different light sources composed of different spectral output.
D50 is so specific in fact, that the only thing that can create it is the sun as it passes through the Earth's atmosphere.
Why choose a standard like D50 as an aim point if nothing man-made can achieve it? Well, because our visual system is based on millions of years of evolution under the sun. We are tuned to see objects illuminated by our star and filtered by our atmosphere.
The light source that I have seen to be the closest to D50 is the Solux lamp designed by Tailored Lighting. It is a Halogen light that can be easily mounted in track-lighting systems and is also quite inexpensive (around $9 US per bulb). The Solux CRI is 98, the industry's best. You can find out more information about these bulbs from solux.net.
If you are not interested in the more "do it yourself" side of lighting that track-lighting sometimes requires, consider GTI or Just Normlicht light boxes. They are fluorescent-based lighting systems which may not have as high a CRI but they are so widely used (especially GTI in the US) that they are de facto standards themselves. If you want to view things as many others view them, light booths will typically get you there.
So, if a company is calling their lighting D50 I'd love to hear about it because it's probably not true. If they are calling it 5000K then they've chosen the broad side of a barn as their target. And if they quote a CRI they may be closer. At least they're more savvy. Remember D50 is the aim-point and so far Solux is the closest.
Myth #18: The "Preserve Color Numbers" checkbox in Photoshop's proof setup doesn't matter much.
Photoshop does some amazing things. Not only will it carefully convert image colors into monitor colors on the fly it can also go through a press profile on the way. This soft-proofing capability is what gives Photoshop the ability to simulate how colors will look if separated using a particular profile OR when sent to a particular device.
This may sound like the same thing but I assure you, it isn't. Let's walk through the soft proofing of two different files to illustrate my point.
First, an RGB file you want to simulate sending to press CMYK. In Photoshop, choose the View: Proof Setup: Custom. For Profile: choose the profile you would use when sending your file to press and for Intent: perceptual or relative colorimetric depending on image content. After clicking OK, Photoshop will soft proof your image to screen. The detailed path your image takes is: working space-> Lab-> press CMYK-> Lab-> monitor RGB. The Lab->CMYK->Lab transformation is what we often call a "round trip" and is responsible for compressing the image gamut and simulating press behavior.
Second, let's soft-proof a CMYK file. Where this CMYK came from doesn't matter too much, we'll soft proof to see how it'll look when sent to press. Open soft proofing the same way as above with View: Proof Setup: Custom. Note, however that this time the "preserve color numbers" checkbox is active. Check it and click OK. You will now see a soft proof simulating how your CMYK file will look when sent to press. The detailed path of the file is CMYK-> Lab-> monitor RGB.
Here's the important part. Had you not checked the "preserve color numbers" checkbox then Photoshop would CONVERT the CMYK file TO your CMYK press profile and then send it to the monitor. Its path would be: working space CMYK-> Lab-> press CMYK-> Lab-> monitor RGB. See the difference? Let's see it again.
PCN on: CMYK-> Lab-> monitor RGB PCN off: CMYK-> Lab-> press CMYK-> Lab-> monitor RGB
That extra step is a big difference. In the first case you are asking "show me this CMYK file as if it were sent directly to press". In the second case you are asking "show me this CMYK file when it is converted to the press profile then sent to press". Unless you are specifically and explicitly wanting to reseparate the file, you want the first case, with preserve color numbers checked.
So, to review: Check "preserve color numbers" whenever you get the chance. Come back and reread this text if you get the urge to uncheck it.
Myth #19: You need a RIP for your printer if you want to proof.
First, what does a RIP do? Originally a RIP was required to rasterize any vector data you wanted to print. Rasterizing is the process of converting lines and curves into dots. All printing uses dots these days so somewhere along the print path the lines and curves in graphics and fonts need to be converted to dots that can actually be printed. Postscript has been a popular language for describing these lines and curves and Postscript RIPs will take complex page descriptions and ready them for dot-based output.
As RIPs evolved more features were added. Now RIPs give greater control over the printer and preprocess incoming files as well. For printer control you can measure density curves and linearize each channel, change ink limiting and so forth. For file preprocessing you can nest images to save time and paper, apply profiles for matching or proofing, manage print queues for multiple users and many other options.
These features are great. They can speed up your workflow, centralize your administration and increase the quality of your work. BUT, they are not required in order to proof.
Proofing, in the color management world, is the act of transforming your colors to your final output profile and then transforming them again to your printer. If you have sufficient control over your printer and you have good profiles, the match between your printer's output and the final destination can be very good.
We build many profiles for inkjet printers that are driven using the manufacturer's RGB-based printer driver. When you control a CMYK printer using RGB you give up the ability to have the profile manipulate the K channel. While this might cause problems in shadows and can make grays challenging, we have great results overall. Coupled with an accurate reference profile, these inkjet profiles can make great proofs. The most important thing in this workflow is the application. Photoshop and other Adobe applications are now smart enough to perform the "proofing transform" that's required for hard proofs. If you are outside of publishing applications then you may not have the ability to insert your reference profile into the print path and so proofing may be out of your grasp. (although you should take a look at our ColorCast technology as it will modify normal printer profiles to proof automatically)
So a RIP, while a powerful tool that can add many capabilities to your printing system, is not absolutely required for proofing. Try it without a RIP first. All those features come at the price of complexity so if you can work without it you've simplified your life a bit. Gives you more time to figure out that new cell phone, camera, iPod, microwave oven...
Myth #20: Adobe Gamma is pretty close to what you can get from an instrument calibration.
Adobe Gamma is a small utility that has been around a long time and until recently was always installed with Photoshop. It is still installed with the Windows version of Photoshop but has been replaced by Apple's default calibrator under OS 9 and X so Adobe stopped distributing it with Photoshop for Mac. Both calibration utilities are "by eye" calibrator software that require no hardware.
So why bother with a hardware device? A few good reasons: white point, gamma, color and repeatability.
When you calibrate your display with a hardware device, a few important functions occur. First, the device measures the colors of the phosphors (CRT) or filters (LCD) of your display. Then it sets (often with your help) the maximum output for each RGB channel, balancing them to obtain the white point you requested (D50, D65, etc). Finally it "ramps" up in each RGB channel and sets the gamma in your graphics card. This fine-tunes the curves for each channel, ensuring that they have smooth color transitions from 0 to 255, compensating for any non-linearities in your graphics card or display, and makes sure that equal RGB values are gray all the way from black to white.
When you calibrate with a "by eye" system it asks you which monitor you have (to estimate the phosphor / filter colors), allows you to choose a white point (sometimes letting you fudge it by moving a slider) and also sets the requested gamma in the graphics card. Sometimes the system can obtain the phosphor/filter colors directly from your display through its cable. The problem is that each of these are an estimate and the errors introduced at each step can compound to greatly reduce the accuracy of the resulting profile.
Let's break these down:
- White Point. You asked for D65. A hardware calibrator will measure your display, adjust the maximum output of the graphics card, and ensure you get D65. "By eye" calibration will set the graphics card at some preset and then hope you get something close to D65. Your display's native white point will play a huge role in affecting the actual white you see and the software has no idea whether it's in the ball park or not. Not good.
- Phosphor / Filter Colors. A hardware calibrator measures these for use in the resulting profile. "By eye" calibration may obtain the numbers from the display or it may ask you for what type of screen you think it is and then substitute numbers it has from a table. This means the software is guessing. The results will range from "OK" to "poor" and will mean the more saturated colors on your display could show noticeable hue shifts (tomato reds , purple blues, etc).
- Gamma. If you want a gamma of 2.2 a hardware calibrator will give it to you. Software will show you a small graphic that requires moving a slider until two elements match in intensity. This software method is better than nothing but nowhere near what hardware can do. This also plays a role in neutrals...
- Gray Neutrals. A hardware calibrator measures each R=G=B combination from 0 to 255 to ensure that they create neutral gray all the way from black to white AND that the transition is as smooth as possible. Your display, graphics card, cables, system software and other pieces all conspire to make curves bumpy. Measuring carefully at each step helps smooth them out.
- Repeatability. No matter how good a job you do with "by eye" calibration, the next time will probably be noticeably different. If you calibrate regularly (you ARE calibrating regularly right!?), hardware calibration will keep your system stable. "By eye" calibration could introduce more fluctuation into your system than if you had left it alone! scary.
Should "by eye" calibration be avoided at all costs? Not necessarily. Sometimes you are out presenting somewhere and want the overhead system to look good. If you don't have a Beamer then a quick run through of software calibration can help. It always seems to improve things for me. But for a desktop system you want to believe, use a hardware device. The price of such devices has come down to the point where it is affordable and should really be the first dollars you spend on color management.
Thanks for reading,
For more Color Management Myths, see Color_Management_Myths_21-25