The following visualizations was created by UCSD undergraduate student Megan O'Rourke in my Winter 2012 class data visualization and compututional art history (the link is to Spring 2012 version of the class).
The visualizations compare color palettes in paintings of of six Impressionist artists. In each image plot, x-axis = median saturation; y-axis = median hue. The visualizations were created with our open source ImagePlot software. Click on each image to see high resolution version.
Data:
Images of 630 Impressionist paintings.
Source:
Artstor.
Number of paintings per artist:
Artstor contains only some of the paintings by these arists. The diffrences in the numbers of images available for each artist reflect the differences in popularity of each artist as well their varied productivity.
When I wrote a macro which we distribute with ImagePlot and included average hue as one the image measurements, I was not sure if it can be actually used in a meaningful way. Megan's visualization shows that this measurement is quite useful. Her other invention was to extend plots vertically and place them side by side - a really good format for comparing image sets.
Teaching of art history for many decades of the 20th century was built on comparing two images at a time using slide projector technology. Media visualization methods developed at our lab allow scaling of comparision. Any number of images can be placed side by side and sorted according to multiple attributes (including metadata, content and visual form).
Close-up: a comparison between Cassat, Monet, Morissot:
Looking at the visualizations, I am struck by how similar are the color footprints of the Impressionist artists. What are the reasons for this similarity? How does it relate to the range of subects in their works, and their habit of working outside using newly available technology of oil paint in tubes?
close-up: all 1226 Impressionist paintings:
