Hue Color Comparisons (2019) | Hacker News Reader

Hue Color Comparisons (2019)

uwdata.github.io

57 pointsthih917 comments

Somewhat similar exercise by XKCD: https://blog.xkcd.com/2010/05/03/color-survey-results/ (2010)

Thank you all for your interests and comments! I'm not sure how the study came here now but glad that more people can see the results :) At that time, we tried to collect many color names as much as possible but the pool was biased as we advertise from our friends/co-workers/websites.

You can contribute to the data by here: http://labinthewild.org/studies/color_perception/?locale=en

Another related article is here: https://idl.cs.washington.edu/papers/multi-lingual-color-nam...

Thanks!!

This highlights an oddity I've always found fascinating about the Russian language: blue. Most languages have blue, then dark and light blue, and some smaller niches for specific niches like turqoise etc. Russian has "regular" blue (синий), a darker version of that blue (темносиний), and a lighter version of that blue (светлосиний). So far so good. But then it also has a very specific "lighter" blue (голубой), a darker "lighter" blue (темноголубой), and a lighter "lighter" blue (светлоголубой). What is it, I wonder, that historically brought forward the existence of two very distinctly different kinds of blue where most other cultures use one?

Then there's Polish:

granatowy (blue, dark)

niebieski (blue like sky)

błękitny (blue, light)

siny (blue like a bruise)

chabrowy (blue like the flower)

The first two are definitely perceived as distinct colors in Polish mind and can be similarly modified with jasno- and ciemno- for lighter and darker hues. The last two are much more rare and like shades especially if you want to communicate more poetically.

"голубой" sometimes not only portrayed as lighter blue RGB(x,255,x), but with some greenish tint RGB(y,255,x) where y > x. But not as much as "marine" tones. :) I also find russian (and some other) are very underrepresented in this study. There are dozens of specific color names, mainly used by painters and designers, but also known to general public.

I feel the data could be more interesting if represented in more perceptually uniform colorspace.

Hi! You may can find some results on Lab space (fairly perceptually uniform color space) from here: https://idl.cs.washington.edu/papers/multi-lingual-color-nam...

We wanted to map more languages on that space but due to the lack of data, we could not. T_T

Too bad there is no Japanese. If a traffic light is green, the Japanese say it is "青い" (aoi), which is usually translated as "blue". I would have loved to see the hue range of this color.

Reader's Digest (of all things) has an article about this that claims their traffic light name is originally an artifact of Japanese words for color but in the 70s the government actually shifted the lights towards blue to match the word. https://www.rd.com/article/heres-japan-blue-traffic-lights/

  The traffic lights, they turn blue tomorrow  
  And shine their emptiness down on my bed  
  The tiny island sags downstream  
  'Cause the life that lived is dead

There's also the problem that people have red-green color blindness, but not red-blue color blindness. Having the light be blue enough that color blind people can tell the difference makes sense.

Same happens in Korean for technically what's green. Most adults will say 'blue light' for signals in a quick response. It really has to do with Chinese characters for blue 청[靑] and green 녹[綠]. When examining the words, they're both describing something that is 푸르다. 푸르다 means something that is related to the clear sky, sea, or the color of grass. Unfortunately, Google Translate just responds that 프르다 simply means blue which it's not.

Also my take: Your tongue twists a bit less when saying blue in Korean rather than green. Can lazy reflexes affect language?

More info here (warning: Korean) https://kidshyundai.tistory.com/990#:~:text=%ED%95%9C%EC%9E%....

青 also means "blue/green" in Chinese. This phenomenon has a Wikipedia article! https://en.wikipedia.org/wiki/Blue%E2%80%93green_distinction...

I was expecting some oddities for Portuguese, given an old chain shift*, but it looks surprisingly normal in comparison with Italian and Spanish.

Also, someone should check the data for ⟨azul-marinho⟩, it's usually a deeper blue than ⟨azul⟩.

*⟨roxo⟩ went from red to purple, while ⟨vermelho⟩ went from an orange-red hue to red.

Very interesting! My wife (US) and I (FR) always disagrees on the color of things, it's amazing to see that it might be a language/culture thing more than just our individual perceptions.

I wonder if there could be a way to build a "personal" spectrum based on our own answers, to compare with another person's. I would be very interested in seeing how the color names vary between me and my wife.

I am also wondering if those spectrum should have both hue and saturation, I am not sure if people consistently name a hue the same across all saturation values (or lightness values).

I set

    * { stroke-opacity: 0 !important }

then tried to see if I could tell where the borders were. For instance, to me the colors in English teal, turquoise, cyan, and aqua all look the same.

You're misreading the chart. It's a histogram of what _names_ people give to a particular color, shown at the bottom of the chart. The regions of color in the chart are just the weighted average color of the samples that were given that name. The chart does not claim to show what the archetypal hew for any given name is, nor does it try and find the boundary between two colors.

This is why in English 'Hot Pink', 'Magenta', and 'Fuchsia' all look about the same ... it's because they're different words for basically the same color.

A similar situation is going on with the blue-greens you named. If you look at the slice of the graph for CSS cyan (0% red, 100% green, 100% blue), you'll see that about 20% of respondents called that color "blue", with light blue, cyan, aqua, teal, and turquoise all taking roughly equal amounts of the remainder. Teal, turquoise, and aqua all take approximately equal portions of the same colors, explaining why their representative averages are nearly identical.

Orange, yellow, and red are very consistent in having almost no subsets.

Reply on news.ycombinator.com