What Color Is This Dress?

By Geoff Walker

Jim Larimer’s Sunday Short Course,The Fundamentals of Color Science and the CIE Color Standards, kicked off with a dress photo that went viral on the Internet in February 2015.  Depending on the observer, the dress appeared to be either blue and black or gold and white. What you can’t tell from the photo is the color temperature of the light in which the photo was taken. It turns out that it was taken at sunset in very reddish light. The point of the photo (and a recurrent theme in Jim’s course) was that color is an experience, not a physical property of material. Everything about color perception is contextual; expectations about the light-source influence the perception of color. For the full story on the dress, see this CNN video and these articles in Wired and in Wikipedia.

In one of dozens of interesting illustrations and illusions, Jim demonstrated that the colors we experience also depend upon the spatial context near the region of interest. He did this by enlarging his consulting company's (Image Metrics) logo. When viewed at small size, the colors of the words “Image” and “Metrics” seem obviously different. Yet when enlarged sufficiently so that the spacial frequency content of the image on the retina is much lower, the colors are actually seen to be identical. The spatial signal evoked by an image depends on both the edges in the scene and the viewing distance.

  Jim’s course was built around three main themes, as follows:

·         Color is an experience, not a physical property of material (as above)

·         The human vision system is an amazingly complex mechanism

·        The 1931 CIE XYZ color system is good for controlling the image created on the retina (i.e., the “quantum catch” signals created in the outer segments of the cone photoreceptors) but it’s not applicable to our perception of color, which depends on the spatial and temporal context of the image and how the quantum catch signals are processed by the human visual system.

It’s impossible to report on even a tiny fraction of the 217 informative slides that Jim presented in his four-hour course. But part of the reward of attending a course that covers the interaction of humans with their environment is the abundance of interesting “factoids” that one picks up along the way. For example:

·      Every organism on Earth that is able to sense light developed that sense based on the spectrum of sunlight, shown in the following illustration from the Sunlight article in Wikipedia.

It would be easy to write another entire page explaining all the information embedded in this illustration; instead, I recommend reading the Wikipedia article. Jim also recommended the book Animal Eyes, which compares all known types of eyes in the animal kingdom.

·      The blood vessels in the human eye cast shadows on the retina, but we don’t see them because they’re not moving.The human vision system is continually calculating derivatives, so if there’s no motion we don’t see anything.

·       No matter how bad our visual acuity is, the world still looks sharp.  This is because the world is a percept (an impression of an object obtained by use of the senses). The signal processing that our brain does in vision is vastly better than any computer that anyone can build yet, but even with as much as we know, it’s still not clear how much processing is happening in the early eye-cells versus in the brain.

·       At the time the CIE color-space was created in 1931, there were no calculators capable of easily handling negative numbers.  The color-matching functions were therefore made all-positive.  This in turn led to the necessity for non-real primaries, i.e., the resulting XYZ values are imaginary primary colors that are outside of the gamut we can actually see and cannot be created with real lights.

·       While industry is struggling to create adequate high-dynamic-range (HDR) cameras, we humans already have it.  We can view a scene that contains both deep shade and sunlight and see detail in both.  The reason is that our retinas can change sensitivity locally, that is, one part of the retina can have a different sensitivity than another part.

·       The tines of a hair comb far enough away will appear to be black and white regardless of what color they actually are.  This is because the maximum spacial frequency that we can see is about five cycles per degree.