A rainbow for everyone

Traffic lights for everyone

Stephen Westland of Colour chat recently posted about a clever new LED traffic light developed in Japan. Here’s my tweet with the link to Westland’s original blog post:

I really like the idea of making a traffic light that works for everyone: for people with full color vision and people with color vision deficiencies. In fact, I think we should do the same with our color palettes. Why do I say that?

A rainbow for everyone

Take a look at  Figure 1 below. This is a map of the Bouguer Gravity (terrain corrected Bouguer Gravity to be precise)  in Southern Tuscany, colored using a rainbow palette. I intentianally left out the colorbar. For a moment ignore the sharp gradient changes at the yellow and cyan color (that is one of the topics of my upcoming series “The rainbow is dead…long live the rainbow!”). Can you tell which color is representing high values an which low? If you have used a mnemonic like ROY B GIV and can tell that highs are towards the Southwest and lows towards the Northeast, then you are right and you also have full color vision, just like me. Great, because this post is exacly for us, the “normals”.

Figure 1

Take now a look at Figure 2:

this is exacly the same map, but I run a simulation to show us normals how it would be seen by a person with Deuteranopia, a form of color vision deficiency (I run the simulation using Vischeck plugin for ImageJ, see the related contents section). Can you tell now which is high and which is low? I certainly can, but I’d be wrong. Using topography as a methafor, what I see is a nice yellow mountain ridge, from which the surface slopes down to shades of brown, and then white, on either side. There’s a nice big blue lake in the the Northeast but as far as I’m concerned there’s also a low in the South.

Figure 2

In my upcoming seires “The rainbow is dead…long live the rainbow!” i will tackle these topics in depth and I will also propose some new rainbow-like color palettes that are better for everyone, no matter how many and how well you see colors. Stay tuned!

RELATED POSTS (MyCarta)

Is Indigo really a colour of the rainbow?

Why is the hue circle circular at all?

RELATED CONTENT (External)

Color Blind Essentials eBook

Vischeck

Vischeck simulations using ImageJ plugin

Vischeck simulations online

17 responses to “A rainbow for everyone

  1. I like grey-scale ‘colour’ bars. The Human eye is more sensitive to shades of grey than colour. Try a comparison and you can clearly see more detail in a grey-scale map than a colour one. It’s also immediately obvious which is high and which is low…

    • Hi Antony

      Thanks for the comment.

      On your first point I disagree when you mention our ability to detect more shades of gray. I think there is a good amount of evidence and scientific literature pointing to the contrary. From all I have read the human eye in normal conditions can discriminate far more color hues (from many hundreds to up to millions) than shades of gray levels (a few dozens), and also gradual variations in color more easily. I’d quote Jay Neitz, a renowned color vision researcher at the Medical College of Wisconsin:”…each of the three standard color-detecting cones in the retina — blue, green and red — can pick up about 100 different gradations of color. But the brain can combine those variations exponentially, so that the average person can distinguish about 1 million different hues”.

      I do agree that grayscale often looks better than the rainbow. I think a large component of this is because of what you mentioned, that some colormaps (the rainbow being one) interfere with the perception of up/dow or stronger/weaker signal. Martin also pointted to this issue in his comment, when he said that the grayscale does not introduce artifacts as hue changes in several color palettes do.

      I have been trying to understand why is that for the last couple of years doing some independent reaserch. It has been shown (for instance in the papers at the bottom) that monotonicity of either luminance or lightness is the most important perceptual quality in a palette. Grayscale has that quality, whereas rainbow does not, that’s I think the critical difference. And indeed grayscale rates far higher on perceptual tests than rainbow (in the second paper). But I think if we could get the lightness monotonicity right in a palette that has color, then we’d have a superior tool than grayscale. In my future series on perceptual rainbows I will propose an alternative colormap that has the best of both worlds, monotonicity of lightness plus color, and test it with model data and geophisical structure maps to show it is superior, not only for viewers with normal vision but also for those with color vision deficiencies.

      Borland, D. and Taylor, R. M. (2007) – Rainbow Color Map (Still) Considered Harmful – IEEE Computer Graphics and Applications, 27 (2), 14-17.
      Rogowitz, B.E. and Kalvin, A.D. (2001) – The “Which Blair project”: a quick visual method for evaluating perceptual color maps – IEEE Proceedings of the conference on Visualization ’01.

  2. Geophysical data from archaeological sites has long been imaged as greyscale to reduce artefacts caused by hue changes and to approximate apparent contrast as something approaching luminance. That’s really important because we’re often dealing with low local contrast, however, treating a greyscale as a model of luminance isn’t entirely correct either due to non-lineararity of visual response.

    Spectrum palettes IMHO should be avoided for anything except T-shirt level imaging! I prefer to optimise palettes to suit the data, within a cross project framework to ensure compatibility and enable comparison.

    • Hi Martin

      Thanks for the comments. What you say about grayscale in archaeological data imaging is very interesting. There is a lot to be learned from other applications or other disciplines but we often turn a blind eye to them.

      As for the spectrum or rainbow, I could not agree more with you, but we’re not learning the lesson. There are some very interesting statistics in a paper by Borland and Taylor: Rainbow Color Map (Still) Considered Harmful (http://www.jwave.vt.edu/~rkriz/Projects/create_color_table/color_07.pdf). They looked at papers from the IEEE Visualization Conference proceedings for the period 2001-2005 and concluded that the use of rainbow in visualization had gone up over time. Also interesting is that the situation got worse when they excluded papers about medical imaging. I find it interesting that the medical (and biomedical) communities have caught up with the idea that rainbow is bad more rapidly and I suspect it is because wrong interpretation due to poor visualization by a doctor has worse consequences than wrong interpretation by a geoscientist. For example, in their paper Evaluation of Artery Visualizations for Heart Disease Diagnosis (http://iis.seas.harvard.edu/papers/2011/borkin11-infoviz.pdf) Borkin et al. argue that using rainbow in artery visualization has a negative impact on task performance, and may cause more heart disease misdiagnoses. I think this goes to show again that we do not pay enough attention to what other scientists do and learn.

      With regards to adapting or optimizing your palette to suit your needs, I really like this article: Task-based Color Scale Design (www.cs.umbc.edu/~rheingan/pubs/scales.pdf.gz). They do use the spectrum in some cases but some of the ideas in the paper are good.
      I agree on the importance of comaptibility and being able to compare.

  3. Pingback: A good divergent color palette for Matlab «·

  4. Pingback: The rainbow is dead…long live the rainbow! – Part 1 «·

  5. Pingback: The rainbow is dead…long live the rainbow! – Part 2: a rainbow puzzle «·

  6. Pingback: The rainbow is dead…long live the rainbow! – Part 1 « MyCarta·

  7. Pingback: Better Palettes « MyCarta·

  8. Pingback: The rainbow is dead…long live the rainbow! – Part 3 « MyCarta·

  9. Pingback: The rainbow is dead…long live the rainbow! – Part 4 – CIE Lab heated body « MyCarta·

  10. Pingback: The rainbow is dead…long live the rainbow! – Part 5 – CIE Lab linear L* rainbow « MyCarta·

  11. Pingback: Comparing color palettes | MyCarta·

  12. Pingback: The rainbow is dead…long live the rainbow! – Part 3 | MyCarta·

  13. Greetings! This is my 1st comment here so I just wanted to give a quick shout out and tell you I truly enjoy reading your posts.
    Can you suggest any other blogs/websites/forums
    that cover the same topics? Thank you!

  14. Pingback: Parula: a new Matlab colormap | MyCarta·

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s