Modeling Color Spaces With Blender 35
Mrs. Grundy writes "When creative professionals want to visualize colors in three dimensions, they often use dedicated and sometimes expensive software. Photographer Mark Meyer shows how, with the help of its Python scripting interface, you can create graphics of color models in Blender. He demonstrates plotting in XYZ, LAB, and xyY space, and also includes the Blender file to show how it's done."
Pretty but why? (Score:4, Interesting)
Why massage and hack a program like blender when you can use the venerable POV-Ray [povray.org], open source raytracer since 25 years back, first raytracer in space, etc.
You can already do all of this directly in its scene description language, and you will get exact results instead of interpolated meshes.
Sort of a flawed premise in the summary... (Score:5, Interesting)
Just for the record, no creative professionals use dedicated and expensive tools to visualize color spaces. If they use an expensive tool like Maya for it, it's because they happen to have it handy for more sensible purposes. Visualizing color spaces is really just a novelty for most people. Anybody who needs to do it regularly isn't so much a "creative" professional, as a color scientist.
Still, sort of a neat demo of the Blender Python API.
Re:Deuteranomaly power :-) (Score:2, Interesting)
Awesome. The amount of misinformation and misunderstanding about anomalous color vision -- even among people who should KNOW better by now -- is staggering. The truth is, the world isn't as neatly RGB as most people believe, not even for chromatypicals (people whose color vision more or less lines up with statistical norms). Read about the history of things like CIE color, and it quickly becomes obvious that they didn't so much come up with a precise definition of colorspace as define the dogma of an RGB religion that most people automatically buy into and assume is a 100% accurate description of how the world works.
What? You think RGB is fine? OK, write a program that displays a black window, and draws 24-bit gradients that run in 512 steps from 000 to f00 to fff for red, and do the same for green (000 to 0f0 to fff), blue (000 to 00f to fff), yellow (000 to ff0 to fff), cyan (000 to 0ff to fff), and magenta (000 to f0f to fff), and tell me they all look smooth and balanced, instead of looking like somebody's drunken crack fantasy. I'll admit that it's worse for me because I'm deuteranomalous, but even for people with NORMAL color vision, it doesn't look the way you'd expect it to look. That's what's wrong with RGB... it's a piss poor compromise for most, and just plain wrong for about 10%.
Take the whole reconciliation of "gamut" with "RGB". There's a HUGE chunk of the spectrum that can't be precisely reproduced with mainstream RGB color, and it's not just because of the difference between pure white and pure black. We capture images with sensors that have a seriously warped view of the universe, then display them on output devices with an even MORE limited ability to accurately reproduce the world. Many of those limits exist because nobody ever questions WHY they're pervasive, and whether we could do better.
Let's take the example of someone with normal trichromatic vision. If you took a camera whose bayer array was laid out to sample light at not just the traditional peak intensities and sensitivity curves of chromatypical red, green, and blue, but ALSO at the deuteranomalous green peak, the protanomalous red peak, and the "lumirod" peak, and made monitors with an additional subpixel color that lined up with deuteranomalous yellow, the resulting image would look enormously better to anomalous deuteranopes and tetrachromatic women, but would ALSO look better to everyone else as well. There would be no need for anomalous protanopes (red-weak) individuals to try turning up the red in a futile attempt to make the monitor's colors look more accurate (something that's as futile as trying to get accurate true-color images out of the Mars Rover Curiosit, for exactly the same reason), because a signal calibrated to the metamers of a tetrachromatic woman would automatically be tuned to the proper me tamers for deuteranomalous & protanomalous individuals as well, and would look either no worse (or slightly better) for everyone else. Well, ok, it wouldn't quite be perfect for protons, but it would still be a hell of a gigantic improvement over the RGB status quo.
Plus, ROYGCB cameras and RYGB displays would give the consumer electronics industry a whole new reason to get everyone to dump everything they own and replace it all, with the perk that since some women would be the major beneficiaries (either because they're tetrachromatic, or because they'd no longer be watching TVs that a deuteranomalous or protanomalous man had to try and tweak to HIS color preferences), it would all have built-in Wife Acceptance Factor.
Re:Deuteranomaly power :-) (Score:3, Interesting)
One of the more fascinating ones to read is this one: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2596756/ [nih.gov] ("Protanomaly without darkened red is deuteranopia with rods").
One thing that makes it so hard to talk about deuteranomalous gamut is the fact that the best and richest part of it mostly lies OUTSIDE the gamut of traditional RGB. Imagine if I gave you a yellow laser, then asked you to mix it with light from a monochromatic orange laser until it matched a red laser. It's impossible -- no matter how dim you make the yellow and how bright you make the orange, it will never look "red". That's the problem deuteranomalous individuals have with everything from photos to video... they always look "wrong" compared to real life (the same way as the Mars Rover's pics), and there's no adjustment we can make to compensate for it... even if we manage to flawlessly match one specific shade of orange by adjusting the relative brightness of red and green, the color of everything ELSE is going to be screwed up even worse. Our color definitions don't match everyone else's, and most of OUR colors get rolled off, attenuated, and mangled away when reproduced in CMYK or RGB.
The best way to illustrate the true spectrum in a way that would preserve everybody's detail and help people with "normal" vision understand deuteranomalous gamut would be to build a row of monochromatic lasers whose frequencies ranged from "pure green" (~500nm, I believe) at the left to extremely intense near-infrared (around 850nm) at the right... with extra lasers at smaller intervals around "orange", since that's where OUR bands of different colors are all clumped and smashed together.