Doing Digital Art History: a data-mining project on open contentPosted: April 10, 2015
Here’s a little data-mining project I absolutely love: charting the colours used in paintings throughout history, by analyzing the pixels of digitized artworks hosted online. (The obvious caveat is to make sure you’re sampling from collections digitized with some fidelity, instead of, say, most of these copies … )
The creator put all of his code in R online, so you could query the exact same collection to do similar analyses with no trouble at all (if you were into that sort of thing).
I made a visualization of the change in colors of paintings over time which a friend tweeted. Several people wanted more info on the method used, so I decided to write a detailed description here, also including the (not very pretty) code I used.
I decided to try to do something similar but with data for a longer time period than the era of film. I decided to download images of paintings. So there is a bunch of different sites where you can access (photos of) paintings, e.g. BBC, Google Art Project, Wikiart, Wikimedia commons, and various museums. One of my favorites is the BBC:s site where you can browse through over 200K of well organized paintings! An amazing resource. For many of these there is also information on the year they were painted, the artist, etc.