Martin Krzywinski / Genome Sciences Center / Martin Krzywinski / Genome Sciences Center / - contact me Martin Krzywinski / Genome Sciences Center / on Twitter Martin Krzywinski / Genome Sciences Center / - Lumondo Photography Martin Krzywinski / Genome Sciences Center / - Pi Art Martin Krzywinski / Genome Sciences Center / - Hilbertonians - Creatures on the Hilbert Curve
Poetry is just the evidence of life. If your life is burning well, poetry is just the ashLeonard Cohenburn somethingmore quotes

information: beautiful

In Silico Flurries: Computing a world of snow. Scientific American. 23 December 2017

Color Resources + Tools

Converting, finding, clustering and naming colors

These color resources are a side project and provided absolutely free to use with no restrictions. If you find them useful, your donation would be a way to say thanks.

Convert colors and white points between color spaces

Download colorconvert
Linux/Win, v0.1 5 Jan 2019

The colorconvert (read documentation) converts colors between color spaces, white points and RGB working spaces.

colorconvert is very useful for analyzing and transforming color coordinates. The output can be easily parsed by downstream scripts or imported into a spreadsheet. You can read colors from a file.

For example, you can use colorconvert to get the RGB and XYZ color coordinates of all white points at each color temperature in the range 4,000–25,000 K.

bin/colorconvert -from 4000K -to RGB255,RGBhex,xyz -oneline
bin/colorconvert -from 4050K -to RGB255,RGBhex,xyz -oneline
bin/colorconvert -from 25000K -to RGB255,RGBhex,xyz -oneline

See the bin/ script.

The tool has support for the following color spaces: RGB XYZ xyY Lab LCHab Luv LCHuv HSL HSV CMY CMYK YCbCr YPbPr YUV YIQ LMS, RGB working spaces: 601, 709, Adobe, Adobe RGB (1998), Apple, Apple RGB, BestRGB, Beta RGB, BruceRGB, CIE, CIE ITU, CIE Rec 601, CIE Rec 709, ColorMatch, DonRGB4, ECI, Ekta Space PS5, NTSC, PAL, PAL/SECAM, ProPhoto, SMPTE, SMPTE-C, WideGamut, sRGB white points: A B C D50 D55 D65 D75 D93 E F11 F2 F7 4000K-25000K.

# convert an RGB color to all color spaces
> bin/colorconvert -from 41,171,226
RGB255       41      171      226
RGBhex   29ABE2
   RGB    0.161    0.671    0.886
   XYZ    0.292    0.351    0.772
   xyY    0.206    0.248    0.351
   Lab   65.815  -15.265  -37.260
 LCHab   65.815   40.266 -112.279
   Luv   65.815  -42.283  -57.421
 LCHuv   65.815   71.309 -126.367
   HSL  197.838    0.761    0.524
   HSV  197.838    0.819    0.886
   CMY    0.839    0.329    0.114
  CMYK    0.725    0.216        0    0.114
 YCbCr  148.295  156.939   98.634
 YPbPr    0.604    0.129   -0.131
   YUV    0.604    0.113   -0.161
   YIQ    0.604   -0.197    0.007
   LMS    0.230    0.410    0.782

# convert a Lab color to RGB255 and RGBhex using Adobe RGB working space
# with single-line CSV output
> bin/colorconvert -from lab,50,-25,50 -to RGB255,RGBhex -oneline -csv -to_rgb Adobe
RGB255 109,128,39 RGBhex 6D8027

# get coordinates of white point for 4000K in RGB255 and RGBhex
# with single-line CSV output
> bin/colorconvert -from 4000K -to RGB255,RGBhex -oneline -csv
RGB255 255,248,187 RGBhex FFF8BB

# convert colors from file
> bin/colorconvert -from spectral.15.txt -to RGBhex -oneline 
spectral-15-div-1 RGBhex 9E0142
spectral-15-div-2 RGBhex D53E4F
spectral-15-div-3 RGBhex D7191C

Snap to colors and find colors

Download colorsnap
Linux/Win, v0.10 6 Jan 2019

The Gretag Macbeth color checker represented in sRGB D65 colors. Colors from RGB coordinates of the Macbeth colorchecker by D. Pascale (download colors).

If you want to find specific colors in an image or snap colors to a set of reference colors then my colorsnap application is for you (read documentation).

This application is useful if you want to figure out what fraction of an image is occupied by a specific color (or color range). The color clustering provided by the colorsummarizer is not useful in this case since the clustering does not key off specific colors. colorsnap will also report the average color of all snapped colors for each reference color and the `\Delta E` of the reference and average.

No shape or position analysis is performed whatsoever. Each pixel is snapped independently.

The application is written in Perl and runs natively on Linux. I've also created a compiled binary for Windows—no need to install Perl. It generates plain-text reports about color proportions, making it perfect for scripting and reports and analyzing the colors in a large number of images.

# man page
> bin/colorsnap -man

# snap tucan image and create snap image and histogram image
> bin/colorsnap -file tucan.jpg -delta_e_max 25 -snap -bar

As usual, for windows replace / in filepaths with \.

For example, the tucan image below was snapped to the 24 Gretag Macbeth colorchecker colors—each color in the original image was matched to the closest color in the colorchecker.

Martin Krzywinski @MKrzywinski
Original image.
Martin Krzywinski @MKrzywinski
Image snapped to nearest Gretag Macbeth colorchecker color.

When snapping to reference colors you can impose a maximum color difference, as measured by `\Delta E`.

Martin Krzywinski @MKrzywinski
Image snapped to nearest Gretag Macbeth colorchecker color within `\Delta E \le 25`. (zoom)
Martin Krzywinski @MKrzywinski
Image snapped to nearest Gretag Macbeth colorchecker color within `\Delta E \le 10`. (zoom)

The application also generates a plain-text report of the color distribution—great for scripting and reports.

      black_2 rgbref  49  49  51 rgbavg  41  45  18 dE  19.7 n 17322  0.105  0.105  0.199 *****                         
         blue rgbref  35  63 147 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
  blue_flower rgbref 130 128 176 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
     blue_sky rgbref  91 122 156 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
 bluish_green rgbref  92 190 172 rgbavg  79 164 145 dE   9.6 n   716  0.004  0.004  0.008                               
         cyan rgbref   0 136 170 rgbavg  50 139 143 dE  17.8 n   139  0.001  0.001  0.002                               
    dark_skin rgbref 116  81  67 rgbavg  85  36  10 dE  22.5 n   655  0.004  0.004  0.008                               
      foliage rgbref  90 108  64 rgbavg  68  83  15 dE  16.5 n 86944  0.529  0.529  1.000 ***********************
        green rgbref  67 149  74 rgbavg  82 136  34 dE  15.9 n  1390  0.008  0.008  0.016                               
   light_skin rgbref 199 147 129 rgbavg 219 150 131 dE   7.8 n    61  0.000  0.000  0.001                               
      magenta rgbref 193  84 151 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
 moderate_red rgbref 198  82  97 rgbavg 176  87  80 dE  13.6 n   490  0.003  0.003  0.006                               
  neutral_3.5 rgbref  82  84  86 rgbavg  73  81  77 dE   5.0 n   268  0.002  0.002  0.003                               
    neutral_5 rgbref 121 121 122 rgbavg 111 120 113 dE   6.1 n    91  0.001  0.001  0.001                               
  neutral_6.5 rgbref 161 163 163 rgbavg 149 164 142 dE  13.3 n    99  0.001  0.001  0.001                               
    neutral_8 rgbref 200 202 202 rgbavg 200 205 171 dE  18.0 n    51  0.000  0.000  0.001                               
       orange rgbref 224 124  47 rgbavg 218  79   6 dE  21.6 n  3853  0.023  0.023  0.044 *                             
orange_yellow rgbref 230 162  39 rgbavg 186 145   4 dE  14.3 n  4466  0.027  0.027  0.051 *                             
   pure_black rgbref   0   0   0 rgbavg  24  17   7 dE   7.9 n  7856  0.048  0.048  0.090 **                            
   pure_white rgbref 255 255 255 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
       purple rgbref  94  58 106 rgbavg  32  42  80 dE  20.8 n     2  0.000  0.000  0.000                               
purplish_blue rgbref  68  91 170 rgbavg   -   -   - dE     - n     0  0.000  0.000  0.000                               
          red rgbref 180  49  57 rgbavg 139  27  18 dE  15.6 n  3512  0.021  0.021  0.040 *                             
    white_9.5 rgbref 245 245 243 rgbavg 251 251 202 dE  23.8 n   371  0.002  0.002  0.004                               
       yellow rgbref 238 198  20 rgbavg 239 217  43 dE  10.1 n 24507  0.149  0.149  0.282 ********                      
 yellow_green rgbref 159 189  63 rgbavg 169 194  36 dE  11.3 n 11707  0.071  0.071  0.135 ****                 

You can uses this application for quick and easy image color analysis. For example, by using the Google maps traffic density colors, you can snap the colors in a Google map to these colors (disregarding all others) to get a sense of the fraction of streets that are busy.

Martin Krzywinski @MKrzywinski
Traffic conditions in downtown Vancouver. (zoom)
Martin Krzywinski @MKrzywinski
Image snapped to traffic density colors within `\Delta E \le 20`. Because some feature icons use a the same or very similar color to traffic conditions, they also appear in the snapped image. (zoom)

   dred rgb 119  39  35 n  203  0.018  0.001  0.028                               
  green rgb 128 211 117 n 7210  0.656  0.026  1.000 ******************************
   lred rgb 224  76  62 n  992  0.090  0.004  0.138 ****                          
 orange rgb 242 155  92 n 2581  0.235  0.009  0.358 **********           

You can use colorsnap to convert artwork to a different palette. For example, below are the colors of the subway lines in New York City, Paris and London.

Colors used by the New York MTA subway lines.

Colors used by the Paris metro lines.

Colors used by the London underground lines.

Colors used by the Tokyo subway lines.

Below is an image of Times Square in New York City snapped to the New York City subway line colors

Martin Krzywinski @MKrzywinski
Times Square in New York City. (zoom)
Martin Krzywinski @MKrzywinski
Times Square in New York City rendered using NYC MTA subway line colors. (zoom)

The Granger rainbow gives a concrete example of how snapping works.

This rainbow is a color calibration image and contains all the RGB colors—here resized as a small image so strictly not all colors are present.

Martin Krzywinski @MKrzywinski
Granger rainbow. (zoom)
Martin Krzywinski @MKrzywinski
Granger rainbow snapped to Gretag Macbeth colorchecker colors. (zoom)

Martin Krzywinski @MKrzywinski
Granger rainbow snapped to subway lines colors from four cities. (zoom)

One way of deriving the reference colors is to use colorsummarizer (see below) to cluster colors in oine image and then using the average cluster color as the input for colorsnap for a different image. Let's try this with the tucan image with Munch's Scream as the reference.

`k=8` means clusters for Munch's Scream (download colors).

`k=32` means clusters for Munch's Scream (download colors).

Above are the swatches for each of the `k=8` and `k=32` clusters of The Scream as analyzed by my colorsummarizer image clustering tool. Below are the colorsnap results for the tucan image using `k=8` cluster colors.

 barley_corn rgbref 185 153  98 rgbavg 110 139  41 dE  34.5 n  3047  0.019  0.019  0.057 *                             
 brown_derby rgbref  88  68  52 rgbavg  57  66  15 dE  24.1 n 34665  0.211  0.211  0.643 ************                  
  brown_grey rgbref 143 131 105 rgbavg  84 116  46 dE  32.2 n  2403  0.015  0.015  0.045                               
 brown_sugar rgbref 142 103  69 rgbavg  82  89  15 dE  29.7 n 53898  0.328  0.328  1.000 ********************          
       cello rgbref  57  80  97 rgbavg  60  76  83 dE   6.7 n    49  0.000  0.000  0.001                               
jungle_green rgbref  40  41  37 rgbavg  36  37  15 dE  12.2 n 25368  0.154  0.154  0.471 *********                     
   limed_ash rgbref  94 102  92 rgbavg  66 127 107 dE  20.8 n  1387  0.008  0.008  0.026                               
     rob_roy rgbref 215 168  77 rgbavg 215 194  34 dE  27.4 n 43683  0.266  0.266  0.810 ****************    
Martin Krzywinski @MKrzywinski
Tucan snapped to `k=8` means clusters of Edvard Munch's Scream. (zoom)
Martin Krzywinski @MKrzywinski
Tucan snapped to `k=8` means clusters of Edvard Munch's Scream. Displayed is the average of the snapped colors for each reference. (zoom)
Martin Krzywinski @MKrzywinski
Tucan snapped to `k=32` means clusters of Edvard Munch's Scream. (zoom)

Color summarizer

The color summarizer generates statistical color summaries of images. Whereas colorsnap (see above) calculates how close the colors in an image match a set of reference colors, colorsummarizer finds `k` such reference colors for which the difference compared to the image colors is minimum.

It reports average RGB, HSV, LAB and LCH color components as well as histograms and individual pixel values for these color spaces. Comes with useful web API for all your automation needs.

Yes! I support LCH, which is extremely useful in generating color ramps and, in general, talking about perceptual aspects of color that are intuitive.

Martin Krzywinski @MKrzywinski
My color summarizer reports the representative colors in an image by grouping colors into clusters of similar colors and reporting the average color in each cluster. This is useful in image identification and comparison.

The color summarizer also identifies representative colors in the image by using k-means clustering to group colors into clusters. The centers of each cluster are also reported by name, using my large database of named colors.

Below is an example of a detailed color report of an image—an adorable Fiat 126p I found while it was screaming out its color against the fading background of Havana.

Martin Krzywinski @MKrzywinski
My color summarizer generates statistical color summaries of images, including a poetic list of words used to describe the colors.

Adobe Swatches for Brewer Palettes

Martin Krzywinski @MKrzywinski
All the Brewer palettes at a glance.

The Brewer color palettes are an excellent source for perceptually uniform color palettes. I provide Adobe Swatches for all colors in the Brewer Palettes.

I also provide a short talk to help you understand why these palettes are important.

Color Palettes for Color Blindness

Color blindness is a thing. You should worry about it when you're designing and especially when you're encoding information.

Martin Krzywinski @MKrzywinski
Sets of representative hues and tones that are indistinguishable to individuals with different kinds of color blindness. The rectangle below the each color pair shows how the colors appear to someone with color blindness.

I provide some background on color blindness and give options for choosing 7-, 12- and 15-color palettes that are colorblind safe.

Color palette for color blindness. / Martin Krzywinski @MKrzywinski
(left) Colors grouped by equivalence of perception in deuteranopes. Each of the two hues is represented in six different brightness and chroma combinations. (right) One of the subsets of colors on the left that are reasonably distinct in both deuteranopia and protanopia. To tritanopes, three of the pairs are difficult to distinguish.

List of Named Colors

Probably the world's largest list of named colors.

With more than 8,300 colors, even a mantis shrimp would be impressed. You can finally imagine a color you can't even imagine and name it!

Martin Krzywinski @MKrzywinski
Use my list of named colors to name the colors in the Google logo: dodger blue, cinnabar, amber and medium emerland green.

The color name list is hooked into the color summarizer's clustering. You can get a list of words, derived from the color names, that describes an image.

Martin Krzywinski @MKrzywinski
The color summarizer returns words that qualitatively describe the image.

color proportions in country flags

A visual survey of the color proportions in flags of 256 countries.

Color proportions in country flags / Martin Krzywinski @MKrzywinski
(right) 256 country flags as concentric circles showing the proportions of each color in the flag. (left) Unique flags sorted by similarity.

Flags are depicted by concentric rings whose thickness is a function of the amount of that color in the flag.

I make the flag color catalog available, as well as similarity scores based on color proportions for each flag pair, so you can run your own analysis.


news + thoughts

Tree of Emotional Life

Sun 17-02-2019

One moment you're :) and the next you're :-.

Make sense of it all with my Tree of Emotional life—a hierarchical account of how we feel.

Martin Krzywinski @MKrzywinski
A section of the Tree of Emotional Life.

Find and snap to colors in an image

Sat 29-12-2018

One of my color tools, the colorsnap application snaps colors in an image to a set of reference colors and reports their proportion.

Below is Times Square rendered using the colors of the MTA subway lines.

Colors used by the New York MTA subway lines.

Martin Krzywinski @MKrzywinski
Times Square in New York City.
Martin Krzywinski @MKrzywinski
Times Square in New York City rendered using colors of the MTA subway lines.
Martin Krzywinski @MKrzywinski
Granger rainbow snapped to subway lines colors from four cities. (zoom)

Take your medicine ... now

Wed 19-12-2018

Drugs could be more effective if taken when the genetic proteins they target are most active.

Design tip: rediscover CMYK primaries.

More of my American Scientific Graphic Science designs

Ruben et al. A database of tissue-specific rhythmically expressed human genes has potential applications in circadian medicine Science Translational Medicine 10 Issue 458, eaat8806.

Predicting with confidence and tolerance

Wed 07-11-2018
I abhor averages. I like the individual case. —J.D. Brandeis.

We focus on the important distinction between confidence intervals, typically used to express uncertainty of a sampling statistic such as the mean and, prediction and tolerance intervals, used to make statements about the next value to be drawn from the population.

Confidence intervals provide coverage of a single point—the population mean—with the assurance that the probability of non-coverage is some acceptable value (e.g. 0.05). On the other hand, prediction and tolerance intervals both give information about typical values from the population and the percentage of the population expected to be in the interval. For example, a tolerance interval can be configured to tell us what fraction of sampled values (e.g. 95%) will fall into an interval some fraction of the time (e.g. 95%).

Martin Krzywinski @MKrzywinski
Nature Methods Points of Significance column: Predicting with confidence and tolerance. (read)

Altman, N. & Krzywinski, M. (2018) Points of significance: Predicting with confidence and tolerance Nature Methods 15:843–844.

Background reading

Krzywinski, M. & Altman, N. (2013) Points of significance: Importance of being uncertain. Nature Methods 10:809–810.

4-day Circos course

Wed 31-10-2018

A 4-day introductory course on genome data parsing and visualization using Circos. Prepared for the Bioinformatics and Genome Analysis course in Institut Pasteur Tunis, Tunis, Tunisia.

Martin Krzywinski @MKrzywinski
Composite of the kinds of images you will learn to make in this course.