San Diego Foursome

CODE: GitHub
DATATYPE: image slices
SOURCE: DigitalGlobe, Google Streetview, Twitter, Flickr
PLACE: downtown San Diego
YEARS: 1994-2016
SIZE: 1 million slices per plot
VIZTYPE: slice histogram multi
FEATURES: saturation, hue
FEAT SRC: scikit-image
BINS: 1440
MORE INFO: Software Studies Blog

This is a slice histogram multi ("large multiples", perhaps). This format is what slice histograms are really made for: comparison. Each of the four plots is a saturation histogram, vertically sorted by hue. Each contains 1 million slices from some number of whole images, a number that varies by image types. In a slice histogram, I aim for uniformity in the average standard deviation of the color properties I'm interested in. After setting some slice size that satisfies this criterion, I then randomly sample 1 million slices (really, 1,048,576) and resize them to 16px on a side. The resulting plots are composed into what we see here. We see in this display that social media adds colors we see rarely in nature — bright, saturated blues and reds.