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
music + dance + projected visualsNosaj Thingmarvel at perfect timing

gotham: fun

Bioinformatics and Genome Analysis Course. Izmir International Biomedicine and Genome Institute, Izmir, Turkey. May 2–14, 2016

visualization + design

ASCII Art—Proportional Spacing, Tone/Structure Mapping and Fixed Strings


  1. ASCII art
  2. proportional spaced fonts
  3. structural character selection
  4. tone-based character selection
  5. fixed string ASCII art
  6. angled text ASCII art
  7. multi-layer ASCII art
  8. recursive ASCII art

download code


This is a Perl script and requires Imager. See README in the archive for instructions. I cannot provide installation support, but welcome questions and ideas about the method.


Martin Krzywinski @MKrzywinski
Part of the Pioneer plaque rendered with the sequence of human chromosome 1, using 8 weights of Gotham. (zoom)
Martin Krzywinski @MKrzywinski
DNA helix rendered with string 'dna'. (zoom)

After finding a typographic portrait of Christopher Hitchens, created out of Gill Sans letters by Miles Chic at Capilano University, I thought to resurrect software I wrote a long time ago that converts images into letters and expanding traditional ASCII art by using

  • proportionally spaced fonts
  • a variety of font weights in a single image
  • both tone and structure of the image to select characters
  • fixed strings to render an image in legible text


The representation of images by characters—ASCII art—has a long history. ASCII art extends the emoticon (or smiley) to represent a larger piece of work. Typically, the works use a fixed-space font (e.g. Courier), originally designed for display on a terminal. Despite the sophistication of computer graphics today, ASCII art continues to have a strong following with new work continually added to public online galleries.

Martin Krzywinski @MKrzywinski
Community contributions to ASCII Art Museum. ASCII art can vary from simple cartoon-like depictions to photorealistic interpretations. (source, zoom)

Photos and paintings can be ASCIIfied using a tone-based approach and automated methods exist to do this (Paul D. O’Grady and Scott T. Rickard (2008) Automatic ASCII Art Conversion of Binary Images Using Non-Negative Constraints).

Many artists generate new creations, exclusive to the medium. Typically this kind of ASCII art is based on the interpretation of structure rather than tone—this method has also been automated (Xuemiao Xu, Linling Zhang, Tien-Tsin Wong (2010) Structure-based ASCII Art).

Proportional spaced and multi-font art

I have written code to generate ASCII art from images by using proportional spaced fonts.

Martin Krzywinski @MKrzywinski
Fixed width fonts (e.g. Pragmata) are popular. ASCII art can be extended to proportionally spaced fonts (e.g. Gotham). More than one weight (or font) can be used to add contrast.

Below is an example of how Pragmata and Gotham can be used to different effect to render an image. When a proportional spaced font is used, the ASCII shape can more fully fill the image.

Martin Krzywinski @MKrzywinski
Comparison of fixed and proportional spaced fonts in ASCII art. Employing multiple weights adds contrast. The grey background is added to emphasize the original image. (zoom)

Let's see how these methods work on a real image. Many ASCII art Mona Lisa versions exist. Below, I render the Mona Lisa with Pragmata, Gotham Book and 8 weights of Gotham.

structural character selection

Two-tone shapes like the S in the figure above require selecting characters that match the structure of the image. (e.g. "|" matches vertical lines). For a given character and image position there are four distinct match possibilities—a combination of whether the character and image have a signal at a position. I show this in the figure below.

Martin Krzywinski @MKrzywinski
Finding the best character involves maximizing overlap (s1, s3) and minimizing penalty (s2, s4).

By maximizing scores derived from matches (s1, s3) and minimizing any penalties (s2, s4), a character is identified based on maximal coverage of the image region and minimum coverage of areas that are blank.

Martin Krzywinski @MKrzywinski
Ink artwork, or thresholded bitmaps in which there are only two tone values, are approximated using structural matching. Here I compare the method of O'Grady and Rickard with my naive structural matching. (zoom)

When proportional text is used, edges are better approximated, such as in the Homer Simpson example below which uses Gotham Book.

Martin Krzywinski @MKrzywinski
For this image, 17pt text matches the detail well. (zoom)

tone-based character selection

Images that are not two-tone require that we match both structure and tone. Structure is approximated by the choice of character, while tone by choice of font weight. To select the best character based on tone, the character's average tone is compared to the average tone of the section of the image to which it is being compared.

Martin Krzywinski @MKrzywinski
Heavier weights are used to match dark areas of the image. (zoom)

It is possible to combine both structure and tone metrics in character selection. Below is an example of how an image with both tone and structure is interpreted as the tone and structure score weights are varied. The balance between these two metrics can be very hard to find—it greatly depends on the image. Tone-based mapping works well when font size is small and the image is viewed from larger distance—in this case, characters play the role of individual pixels with varying brightness. Structure-based mapping works with larger type and closer viewing distance.

Martin Krzywinski @MKrzywinski
A tone:structure ratio of 1:0.5 works well for the Star Trek logo. (zoom)

Continuous tone bitmaps are an idea application of multi-font ASCII art—images no longer need to be thresholded or dithered.

Martin Krzywinski @MKrzywinski
Applying both tone and structure character selection metrics to a greyscale image. (source, zoom, )

fixed string ASCII art

ASCII art is generated by dividing the image into a grid and finding the letter (the choice of characters is often expanded to include punctuation) that best matches the grid section. Typically, for each grid the entire set of allowable characters is sampled. Instead, we can limit the choice of character by successively sampling from a fixed string.

Martin Krzywinski @MKrzywinski
Fixed string ASCII art limits the choice of characters available at each grid. Characters can be drawn from a short string (e.g. 'ilovegotham') or from a larger corpus (e.g. Wikipedia entry for Mona Lisa). The string can be contiguous within the image, or locally within the font. (zoom)

rendered with the fixed string "monalisa" using 8 weights of Gotham.

Martin Krzywinski @MKrzywinski
Fixed string ASCII art rendering of Mona Lisa. (zoom)
Martin Krzywinski @MKrzywinski
DNA helix rendered with string 'dna'. (zoom)
Martin Krzywinski @MKrzywinski
DNA helix rendered with sequence from human chromosome 1. (zoom)

angled text ASCII art

Things get even more interesting when the text is angled.

Martin Krzywinski @MKrzywinski
By applying rotations to the input and output images, the image can be approximated by angled text. (zoom)

multi-layer ASCII art

The image can be textured with multiple layers of ASCII art. In the example below, four layers of text are used, each with a different font size.

Martin Krzywinski @MKrzywinski
Part of the Pioneer plaque rendered with the sequence of human chromosome 1, using 4 layers of sizes (17pt, 33pt, 59pt and 93pt) and 8 weights of Gotham. (zoom)

Instead of varying size, the angle of the text can be changed among layers. This results in a pattern reminiscent of a halftone.

Martin Krzywinski @MKrzywinski
Part of the Pioneer plaque rendered with the sequence of human chromosome 1, using 4 layers with different text rotation (-45, -15, 15, 45 degrees) and 8 weights of Gotham. (zoom)

recursive ASCII art

An image can be asciified several times, with each iteration the asciified output of the previous step used as input for the next. At each step, the font size should be reduced to s → √s.

Martin Krzywinski @MKrzywinski
ASCII, set in Gotham Ultra (zoom)
Martin Krzywinski @MKrzywinski
The above image asciified using 8 weights of 105pt Gotham and the fixed string 'ASCII'. (zoom)
Martin Krzywinski @MKrzywinski
The asciified example above, asciified again using 8 weights of 11pt Gotham and the fixed string 'ASCII'. (zoom)

news + thoughts

Unentangling complex plots

Fri 10-07-2015

The Points of Significance column is on vacation this month.

Meanwhile, we're showing you how to manage small multiple plots in the Points of View column Unentangling Complex Plots.

Martin Krzywinski @MKrzywinski
Nature Methods Points of View column: Unentangling complex plots. (download, more about Points of View)

Data in small multiples can vary in range, noise level and trend. Gregor McInerny and myself show you how you can deal with this by cropped and scaling the multiples to a different range to emphasize relative changes while preserving the context of the full data range to show absolute changes.

McInerny, G. & Krzywinski, M. (2015) Points of View: Unentangling complex plots. Nature Methods 12:591.

...more about the Points of View column

Fixing Jurassic World science visualizations

Fri 10-07-2015

The Jurassic World Creation Lab webpage shows you how one might create a dinosaur from a sample of DNA. First extract, sequence, assemble and fill in the gaps in the DNA and then incubate in an egg and wait.

Martin Krzywinski @MKrzywinski
We can't get dinosaur genomics right, but we can get it less wrong. (a) Corn genome used in Jurassic World Creation Lab website. Image is from the Science publication B73 Maize Genome: Complexity, Diversity, and Dynamics. Photo and composite by Universal Studios and Amblin Entertainment. (b) Random data on 8 chromosomes from chicken genome resized to triceratops genome size (3.2 Gb). Image by Martin Krzywinski. (c) Actual genome data for lizard genome, UCSC anoCar2.0, May 2010. Image by Martin Krzywinski. Triceratops outline in (b,c) from wikipedia.

With enough time, you'll grow your own brand new dinosaur. Or a stalk of corn ... with more teeth.

What went wrong? Let me explain.

Martin Krzywinski @MKrzywinski
Corn World: Teeth on the Cob.

Printing Genomes

Tue 07-07-2015

You've seen bound volumes of printouts of the human reference genome. But what if at the Genome Sciences Center we wanted to print everything we sequence today?

Martin Krzywinski @MKrzywinski
Curiously, printing is 44 times as expensive as sequencing. (details)

Gene Volume Control

Thu 11-06-2015

I was commissioned by Scientific American to create an information graphic based on Figure 9 in the landmark Nature Integrative analysis of 111 reference human epigenomes paper.

The original figure details the relationships between more than 100 sequenced epigenomes and genetic traits, including disease like Crohn's and Alzheimer's. These relationships were shown as a heatmap in which the epigenome-trait cell depicted the P value associated with tissue-specific H3K4me1 epigenetic modification in regions of the genome associated with the trait.

Martin Krzywinski @MKrzywinski
Figure 9 from Integrative analysis of 111 reference human epigenomes (Nature (2015) 518 317–330). (details)

As much as I distrust network diagrams, in this case this was the right way to show the data. The network was meticulously laid out by hand to draw attention to the layered groups of diseases of traits.

Martin Krzywinski @MKrzywinski
Network diagram redesign of the heatmap for a select set of traits. Only relationships with –log P > 3.9 are displayed. Appears on Graphic Science page in June 2015 issue of Scientific American. (details)

This was my second information graphic for the Graphic Science page. Last year, I illustrated the extent of differences in the gene sequence of humans, Denisovans, chimps and gorillas.