Visualizing the Rijksmuseum

In a Data Visualization course at Carnegie Mellon, for the final project, we were to find and visualize a dataset using d3.js. Our team was interested in the shift in how people were experiencing art as new initiatives spearheaded its digitization. We were inspired by innovative cultural projects, like those of Google’s Arts & Culture. So we chose to use The Rijksmuseum’s newly released API, giving us full digital access to its collection in Amsterdam.

We submitted our project to IEEE Visualization Conference, an annual conference on information visualization. We were selected to share our work in a presentation, poster, and public art installation for the Vis Arts Program, alongside famous information designers such as Giorgia Lupi.

Goal: How can users experience the Rijksmuseum’s art remotely in new ways that are not available through a museum visit? Our project encompassed 4 virtual “exhibits,” or attempts to answer that question.

We believed plotting all of the 4,000 paintings on a timeline, ordered by creation date, would provide a feel for the identity of the museum. It revealed a high concentration of art spanning the 17th century, when The Netherlands experienced The Dutch Golden Age. At this time, Dutch trade, science, military, and art were among the most acclaimed in the world. The period during The Hague School (1860-1890) also had denser coverage. We also wanted to show if a painting was currently on display or in storage. This “exhibit,” a graph, illustrates that visiting a museum can just scratch the surface of the full collection.

We were interested in the use of color in paintings from The Hague School. We read that over time, the artists transitioned their palettes from gloomy and gray to lighter and brighter, under the influence of Impressionism. To dig into this claim, we pulled a representative color sample of pixels from each painting and sorted them by hue, saturation, value, and combinations of each. We found hue to be the clearest. Once we summarized every painting into a strip of its colors, we ordered them by creation date. (The resulting timeline did not support the original claim, but a larger dataset would be needed to verify.) To add another dimension to this “exhibit,” we wanted to help users to get a feel for the physical scale of the works. Using small multiples, we showed the size of the canvases that each artist chose to work with. We used hover interactions to link the individual visualizations together.

We also wanted to explore new, delightful ways for users to traverse a collection. Why not through the characters featured in the paintings? Using the format established by social media, this “exhibit” enabled users to engage through a web of personalities. We ran facial recognition APIs on the paintings, testing a few options. Kairos most effectively identified the faces. It also estimated other characteristics, which we hoped would playfully augment the dataset further.

For the final “exhibit,” we introduced gamification, focusing on the paintings that the Rijksmuseum designated as masterpieces. As many of the paintings are well-known, we wanted to encourage users to guess the title and artist of the work as it’s slowly revealed. Colorful pixels from a work gradually make their way to their respective places on the final “canvas.” We have the pixels flow along the canals of a vectorized map of Amsterdam, to give a sense of place.