End Run


Yesterday a good friend and colorist had this to say:

pJuan is speaking to something to important: the term “science” is loaded with the notion of impartial, objective truth about the universe:


via XKCD No, thank you, XKCD.

Sure, you need to understand some of the underlying scientific principles of color theory in order to calibrate a monitor, design a workflow, or build a LUT.

The problem arises when (probably) well-meaning people start talking about “their” color science, or “ARRI’s color science” vs. “RED’s color science.” Or worse yet, when a post production Dojo breaks out the scary phrases like “Proprietary Color Science.”

At best, this is laziness.

At worst, it’s a shoddy attempt to Blind You With Color Science.

What’s getting confused here is the practice of the scientific method (induction, falsifiable hypotheses) and the application of science (engineering, aka building stuff.)

An aeronautical engineer sure needs to have studied the sciences. But when he builds an airplane he calls it an airplane. He does not call it “A Science.”

i made a science

Color Science is real, and uniquely difficult.

This is not say that color science isn’t a legitimate field of inquiry. Quite the opposite — it is, in fact, a uniquely difficult one. It doesn’t merely concern itself with objectively measurable optical phenomena. Very specifically, it is concerned with the human perception of those phenomena. The discipline of color science therefore overlaps physics, chemistry, biology, and psychology. Which is why we need constructs like “The Standard Observer”.

As one reader pointed out to me: “Most real color science is not even related to the film industry.” That is true. At the same time, many film industry professionals have made significant contributions to the advance of color science. People like Joseph Slomka, Josh Pines, and of course Charles Poynton.

I’ve often been asked why color encoding standards are not neater, tidier, more symmetrical. That one’s easy: we humans are not neat, tidy, or symmetrical. And the models that go around encoding that light and color information have to interact meaningfully with our complex, messy, and evolved human visual system.

Science vs. Engineering

So the problem arises not with the existence of the term color science, but with its casual hijacking.

Chatter about “my color science” or “their science” usually comes down to one party’s specific, subjective approach to a challenging engineering problem that involves color theory. How do I process this particular film stock? How do I build a perceptual transform from P3 to 709 color space? How do I algorithmically convert Standard Dynamic Range footage into a High Dynamic Range image that is pleasing to look at? How to de-mosaic this particular camera’s RAW sensor data?

These are difficult and interesting problems to solve. They all involve knowledge of color theory. But one party’s preferred approach should really not be described as their science. It is merely their technique, their algorithm, their solution, or their product.


xkcd purity

Quick refresher via XKCD

So if you ever meet a post production supervisor whose blood pressure spikes when she hears the phrase “proprietary color science” — hire her on the spot.

UPDATE: This article was revised to better clarify the legitimacy of color science as a field of inquiry, as delineated from the “casual hijacking” of the term itself.

I build Endcrawl and work at HBO. Sometimes I'm wrong about things on the Internet. Feel free to point some of that out on Twitter, or down in the comments section.

  • Sean

    I don’t like this post. Here’s why.

    Let’s start with your friends tweet. Him being a colorist, I have to interpret his use of the term “color science” to actually mean “color management”. These being two interrelated but different things. It is clear that in the film industry the term “color science” or “color scientist”, is just that, an industry term. There are many industry terms within the film world, “film” itself being one of them. So when people use phrases like “ARRI’s color science” the cunning color savant would substitute in “ARRI’s color management”; their engineering solution.

    You do address this issue, but in an odd way. You chose to replace the term “color science” with “color theory”. Which I believe to be an even worse error. Color theory already has well established meaning. To quote Wikipedia, Color Theory is the “body of practical guidance to color mixing and the visual effects of a specific color combination”, which has very little to do with color management. Me being very careful to explicitly interpret color theory traditionally as relating to concepts of color harmony, complimentary colors, or color warmness/coolness.

    Though, in part, I whole-heartedly agree with you. Companies should not be touting their color management solution as their “color science”, but would be equally remiss to start calling it their “color theory” (I think we can both agree there).

    These next points are what irritate me the most, and work to build a tone around your post which I believe is unfounded. Returning to your friends tweet that started this article. If I were to interpret this tweet as not meaning an engineered color management solution, but rather the academic field it’s self, then I would have to agree with him. 80% science ( science interpreted as: explored, well understood concepts, mathematical models, well proven hypothesis’), 15% wishful thinking (areas that aren’t well understood, research areas that are earning people PhD’s), and 5% ¯_(ツ)_/¯. Yup, well doesn’t that just about sound like the case for any scientific discipline? The last 20% being why scientists wake up in the morning and get into the lab. I don’t see this as a dirty secret.

    Nearing the end of your post you then bring up the fact that Color Science doesn’t “live squarely in the domains of math and physics”. Yes, you are correct there, but you raise it as if it were somehow lesser because of that? I don’t see how using a statistical model of a given population dethrones Color Science and adds to the pile of “dirty secrets”. I know you didn’t outrightly state that, but your post entitled “blinded by color science” certainly paints that picture. Additionally, for the interested reader, I suggest looking at the history of Psychophysics of which Color Science belongs, and is in my opinion one of the most interesting areas of science. The handshake between the physical world and the psychical.

    So here’s the points I wish you had actually addressed, and not thrown punches at a scientific discipline of which you don’t belong to. First, in response to your friends tweet, I would say “hahaha” . Second, in an article post, I would talk about the sociological factors in play with modern culture and the ever present scientism of marketing practices. Focusing on the realm of motion pictures and television where it would seem there is an increasing trend towards using “science” incorrectly, especially with color science. There are VERY few people in the industry that are qualified to discuss topics of color science intelligently, and not use it as a cloak to disguise incompetence or misunderstanding. In short, if your local “color scientist” just walks around saying “science, science, science”, and can’t actually explain it over a beer, you should get a refund.

    Your friendly wandering color wizard

    • Chris Healer

      I agree, and I appreciate the term “color management” to distinguish what post production is doing.

      “Color science” sounds fancy and answers to the mystery surrounding a bunch of checkboxes and selection menus in software, but really the “science” part has been done by the time a file is on a disk being read by an application.

      Sean, I think you should dull the point a little in your response, though. In a world of sRGB where lots of things are automatic, it doesn’t remove anything from the smart color scientist to acknowledge a little “if you know what I mean” in order to keep things moving.

      I don’t love sRGB, but it has its place. There are basically 5 color models, and each has a reason for being. But in post production, that’s not very interesting. What is interesting is what gets you out of the murk and home at night.

      I have to belabor one thing which is so often poorly named and under-mentioned by articles and conflated with “color science” or worse “color space”, which is simple gamma curvature. Gamma transforms are how dynamic range is achieved, and what we’re really talking about with sRGB, rec709, sLog, Cineon, or any of about 20 more members of this category. They aren’t color spaces or even color models, they’re just the transform used to get data off a disk and onto a monitor and into a processing space for software to do something useful.

      This also doesn’t take anything from the color scientist who is contemplating metamerism or tetrachromes.

      But when you get the transform wrong, it looks like you don’t know what you’re doing. When Red doesn’t publish their RedLogFilm curve as anything but a .cube file, which most softwares don’t invert, it interrupts the workflow if you need that inversion.

      This is the 15+5% that Juan mentions…. And the odd thing is that the best colorists seem to be the least scientific ones. They just feel it, “if you know what I mean”.

      I also want terms to be complete, accurate, and meaningful for color management. Don’t throw around the term “color theory”, because there really isn’t a theory about it. Within human perception things are very well-defined.

      The article is calling out the mystery part of the equation. To boil it down: light enters a camera, data is saved on a disk. Someone reads that data and does something with it. That data ends up on a screen somewhere. Lots of parts of this equation are well-known. But not all of them. It seems science-y but with creative parts.

Next Post