Wednesday, January 24, 2007

Meet me at EI

The IS&T/SPIE  19th Annual Symposium on Electronic Imaging Science and Technology in San Jose is taking place from 28 January to 1 February 2007. You can find all the information you need at http://www.electronicimaging.org/. I will present two papers, one on digital publishing and one on psychophysics experiments.

The first paper will be in the Color Imaging XII: Processing, Hardcopy, and Applications conference Wednesday 31 January at 10:40 am in Convention Center Room A2. Titled Adaptive color artwork, a pre-print is available at http://www.hpl.hp.com/techreports/2006/HPL-2006-186.html. Here is the abstract: The words in a document are often supported, illustrated, and enriched by visuals. When color is used, some of it is used to define the document’s identity and is therefore strictly controlled in the design process. The result of this design process is a “color specification sheet,” which must be created for every background color. While in traditional publishing there are only a few backgrounds, in variable data publishing a larger number of backgrounds can be used. We present an algorithm that nudges the colors in a visual to be distinct from a background while preserving the visual’s general color character.

The second paper will be in the Image Quality and System Performance IV conference Wednesday 31 January at 1:35 pm in Convention Center Room C2. Titled Web-based versus controlled-environment psychophysics experiments, it is a collaboration with the University of Milano-Bicocca and the pre-print is available at http://www.hpl.hp.com/techreports/2006/HPL-2006-187.html. Here is the abstract: A recent trend in psychophysics experiments related to image quality is to perform the experiments on the World Wide Web with a large number of observers instead of in a laboratory under controlled conditions. This method assumes that the large number of participants involved in a Web investigation “averages out” the parameters that the experiments would require to keep fixed in the same experiment performed, following a traditional approach, under controlled conditions. In this paper we present the results of two experiments we have conducted to assess the minimum value of color contrast to ensure readability. The first experiment was performed in a controlled environment, the second on the Web. The result emerging from the statistical data analysis is that the Web experiment yields the same conclusions as the experiment done in the laboratory.

The colors are wrong

When I meet new people, after they read my business card they often exclaim "Oh, nice to meet you, I have one of your printers and the colors are not the same as on the screen; what should I do?"

Unfortunately, there is no answer that fits on the back of a business card, except maybe a URL which you could also find by using your favorite search engine. In the following I will try to explain briefly why the answer does not fit on the back of a business card.

Color does not exist in nature; it is an illusion that is elicited in our visual system. Therefore, color reproduction is the art and science of predicting an illusion by doing physical experiments, and this is what we color scientists here at HP Labs do for a living.

Our life is made easier by the fact that people do not want us to produce the same color of their original, they just want a color match. This is where it starts to get tricky, because colors match only under certain reference conditions.

Let me make a comparison with my car. Under reference conditions, it makes 38 miles per gallon, but in reality I never reach 38 miles on a gallon. The reason is that I do not drive under the reference conditions, because most of the time in my 2.4 mile commute is spent idling with a cold engine on red lights and stop signs; moreover I drive uphill to work.

In our industry we have it a little easier than in the car industry, because our reference conditions are more realistic in modeling what the actual user has (or at least this was so when the sRGB standard was written). Today, all vendors do an excellent job at matching colors under the reference conditions, and they even do a very good job at preventing things from falling apart when the actual viewing conditions are somewhat off from the reference conditions.

Let us do another comparison with cars. Given my driving habits, I bought a basic car with a small engine and hand-crank windows. Every September I drop it off at my mechanic for a service, which essentially consists in changing the oil, tuning up the engine, and setting the tire pressure.

In the case of displays and printers, the tune-up is called calibration, and for devices in the equivalent range of my car the service interval is so long that you actually forget about it.

Motoring enthusiasts do not buy cars like mine, they buy high performance cars; often they even race. In a race, they do not set the tire pressure; they determine the temperature and road conditions and then choose the appropriate tire. They are also very picky on the oil they use and constantly tune up their cars.

Similarly, if you are a serious photographer, you select the best printing paper for the mood of your image and you keep recalibrating your devices.

Depending on your needs, your calibration efforts will be different. At the "lower end" you can get by just adjusting your viewing conditions so they are close to the sRGB conditions, which are:

Reference Display Conditions

  • luminance level: 80 candela per square meter
  • white point: D65
  • gamma: 2.2

Reference Viewing Conditions

  • screen background: 20% of reference display area
  • surround: 20% of reference ambient illuminance level
  • proximal field: 20% of reference display luminance level
  • ambient luminance level: 64 lux
  • ambient white point: D50
  • veiling glare: 1%

Your device manual may have instructions on how to set it up, or your imaging software may have instructions, including software tools that allow you to do the calibration visually, without having to buy color measurement instruments. You can also search the Web with phrases like "color calibration" and "color workflow".

At the "high end" you will have to bite the bullet and learn all about device calibration, ICC profiles, work spaces, and spectrophotometers. The fastest way is to take one of the many classes. You may also want to follow a blog addressing professional photographers, such as HP's http://www.communities.hp.com/online/blogs/graphicarts/default.aspx.

If you found a Web site that was particularly helpful for you, use the comment feature of this blog to share its URL and your experience.

Thursday, January 4, 2007

Welcome to the "mostly color perception" blog!

This blog is a view from my window — a collection of tidbits I judged relevant to computational color science and in general to the promotion of scientific excellence in areas of major strategic importance for the future of research, economy and society.

I started working in color science in 1986. While performing research in VLSI design tools (my background is in computational geometry), I noticed designers were often making very stupid design errors because they could not decipher well the geometry they were laying out. In the quest of finding a methodology to color-encode layout, I learned color science and radically improved the efficiency of VLSI designers. If you have access to a good library, you may be able to find PARC technical report EDL-88-3 "A New Approach to Imaging IC Layout and Schematics."

A lot of water has flown under the bridge since 1986, but as of 2006 I am back working in a color science project. As part of my work I keep an eye on research in color perception and when I find it of general interest, for example because the result is unexpected, I will post a pointer in this blog.

Having been in this field for many years and having been successful at it, I also have a number of "overhead tasks" for the community, which generally have to do with the governance of research, such as editing journal papers, reviewing and overseeing grants, making recommendations for honours, awards, and prizes, etc. From time to time I will also comment on these topics.

Blog Statement

The Internet is an amalgam of forms blurred under epistemological pressures. In Søren Kierkegaard’s words, under this flat shower of leveled information, where everybody is interested in everything and nothing is too trivial or too important, people just accumulate information and postpone decisions indefinitely, i.e., nobody takes action and nobody is responsible for truth — there is no mastery, just gossip. He called this the æsthetic sphere of existence, exhorting us to evolve to the ethical sphere, where we do not just accumulate information but take action and make commitments. Blogs are instruments to overcome flatness by creating opportunities for vertical activities. In this sense this blog is a view from my window — a collection of tidbits I judged relevant to computational color science and in general to the promotion of scientific excellence in areas of strategic importance for the future of research, economy and society.