BlakeTint is a Yale Digital Humanities Lab project supporting analysis of William Blake’s use of color across multiple editions of his illuminated manuscripts.
This project had two phases, and I served as the primary developer on each.
In the first phase, I wrote a data processing pipeline to quantize images — to “delete” background pixels and generate 20-color summary palettes. This pipeline is available as a suite of modular Jupyter notebooks (enabled by my modular notebook software, Margo), which may be be run online without installation.
The second phase of this project was building a React-based web application with the following views: an index page that renders the entire collection of (4,000+) palettes; and a print viewer page that compares different editions of the same print.
In this view, the objective was to be able to observe macro-level trends in color uses, such as similarities within an edition of a book, or changes over time. Each palette represents a single print. This both compresses the work of Blake visually into a “scannable” and abstracts it, creating a high-level view of the artist and poet’s use of color.
Detail view, comparison slider mode
In this mode, a user can compare different editions of prints to observe minute differences in color use. Blake didn’t just mass produce prints; he augmented them with hand coloration that makes each edition quite unique.
Detail view, quantization mode
In this mode, a user can view an image as the computer “sees” it. The image is recreated using only the 20-color palette, and is shown with the background pixels “deleted.”
- Yale DHLab project page for BlakeTint - https://dhlab.yale.edu/projects/blaketint/
- Color extraction methodology repository - https://github.com/jakekara/color-extraction-methodology
- Color extraction methodology live demo - https://mybinder.org/v2/zenodo/10.5281/zenodo.4554402/
- Summarizing Blake (slide deck) -https://docs.google.com/…