Week 1: Reverse Engineering “Robots Reading Vogue”

Introduction

“Robots Reading Vogue” is a project that was formed by Peter Leonard and Lindsay King from the Digital Humanities department at Yale University that brings together fashion and data mining in hopes to understand various subjects such as gender studies and computer science. Despite Vogue’s signature and classic covers, this project shows that even this influential magazine experienced many changes throughout the century varying in style, color usage, and front cover preference.

Sources

The sources in this project were mainly Vogue covers from 1892 to 2017 along with Vogue Archives from ProQuest LLC. They collected 2,700 covers, 400,000 pages, and 6TB of data. With these resources, they were able to compile a project that correlated fashion and data mining.

Process

Because of the abundant amount of data, in order for it to not be overwhelming each source had to be organized and arranged into subcategories that seemed fit to the research. They separated the covers by date, hand-aligned the covers, arranged the word usage, along with sorted through advertisement and the statistic of each magazine. This lead to different experiments such as Slice Histograms, Cover Averages, n-gram Search, Topic Modeling, Advertisement, Statistics, Student Work, Colormetric Space, Fabricspace, along with Take a Memo.

Presentation

Leonard and King presented the project through a site that is user friendly. They organized the project by experiments and made sure to include a featured image that included a short and simple caption to help give a general idea of what each experiment was about. Two experiments that stood out were the Colormetric Space experiment and the Cover Average experiment.

The Colormetric Space experiment mainly used a tool called ImagePlot which plotted Vogue covers based on how colorful it was. Through ImagePlot, they were able to present a data that was collected from the 1890s to the 2010s.

With this graph, the higher the cover is on the y-axis, the more colorful it was. Throughout the years, the covers significantly got brighter and the saturation increased especially during the 1980s.

Going along with Colormetric Space, Cover Averages was an experiment that illustrated the visual continuity and changes throughout the decades. Their presentation focused on overlaying pictures and generating a mean RGB value for every pixel of an image. 

Even though there are multiple layers of magazine covers overlapping one another, we are able to see the general trend which is that as years go by, the covers become more saturated, brighter, and colorful. Additionally, these images help determine whether the covers are similar and have a clear pattern or if it is completely different and diverse. For example, in the 1980s, the Vogue covers were fairly uniform and consisted of a woman’s face and in the 1940s, we are unable to see a clear visual.