RothkoViz is an online visualization program that organizes 201 images of paintings by Mark Rothko. It offers three ways of exploring data: 2D image plots, 3D interactive spaces, and some image generators.
The first visualization method is presented as simple large static images. One of them localizes crossings on the Cartesian plane that summarize 50+ visual features. The dimension reduction is achieved with a Principal Component Analysis (PCA) calculation. The other image shows an orthogonal view of our archive (which means cropping every image into thin slices), but this time in a polar Cartesian plane. Statistic operations were made with Mondrian and image plots with ImageJ. The expected viewer reactions include understanding image transformations and experimental spatial representations of the same archive that might highlight visual patterns.
The second manner uses extensively WebGL as a graphics engine to explore and navigate the corpus. I first plotted chromatic values as color points with P5.js. The extraction of red, green, and blue values was made with image processing tools included in ImageJ. Then I plotted those values following the RGB color model (a cube whose vertices stand for black, white, red, green, blue, yellow, cyan, magenta). While the first plot rightly depicted color values, I was more interested in displaying the images themselves. For this reason I adapted code with the three.js library to create a virtual environment. After a user session, she could orbit within a virtual world of images floating around the space.
Technically speaking, data was obtained from The National Gallery of Art website (http://www.nga.gov/, Washington, DC). Data was extracted from the web with free tools such as import.io. Visual features extraction and quantification was made with ImageJ and Cultural Analytics Lab tools. Data was cleaned and processed with several tools: MS Excel, Google Spreadsheets, Open Refine, Mondrian. Web development was made with Atom editor.
The methods and lessons learned from the project are used regularly in higher education courses on data visualization. Moreover, it inspired recently an exhibition at the Winchester School of Arts, as a curatorial approach to art and data. https://anti-materia.org/a-curatorial-approach-to-dataviz-eng
Everardo Reyes earned his PhD in Information and Communication Sciences from Université Paris 8 – Vincennes-Saint-Denis in 2007, where he is Associate professor in Digital Humanities at the School of Mathematics, Computer Science and Information and Communication Sciences since 2015. His research areas include visual semiotics, software studies, and media art. He is member of the Paragraphe Lab. at Université Paris 8, the Cultural Analytics Lab. (in USA, led by Lev Manovich), and member of the bureau of the International Association for Visual Semiotics (directed by Göran Sonesson). He has published several books, translations, and congress proceedings, among others: The image-interface (Wiley-ISTE, 2017); Designing interactive hypermedia systems (ed) (Wiley-ISTE, 2017); Archiving and questioning immateriality: proceedings of the 5th Computer Art Congress (ed) (Europia, 2016); Inventar el futuro (Spanish translation of Edward Shanken’s Survey of AEM, 2014). He has organized congresses and curated exhibitions related to media art including the work of Roy Ascott, Jack Stenner, Shulea Cheang, Paul Magee, Christa Sommerer, Malu Fragoso, Pierre Boulanger, Arcángel Constantini, and a panel commemorating the 40th anniversary of IAST Leonardo. More info at http://ereyes.net