Visualizing Color Palettes from Paintings

Presenter: Tanvi Modi

Research Category: Interdisciplinary Topics, Centers and Institutes
College: College of Arts Media & Design
Major(s): Art and Design
Graduation Date: 2022
Additional Authors: Yuhang Li

The Task: Redesign an existing piece of visualization in a creative and insightful way. I chose to redesign Arthur Buxton’s work “Ten Artists, Ten Years” poster into an interactive website prototype with more flexibility in visualization the color palettes of all paintings.

The project involved 3 main sections:

  • Color Extraction: Extracted HSV color pixels from paintings using k-means clustering in Python.
  • Data Viz: Mapped the top-7-8 color pixels into treemaps & bar charts and made 2D/3D scatter plots with the RGB values.
  • Prototype: Created a website dashboard prototype which enables a user to analyse, filter and compare artists by their use of color palettes in paintings.

This project was an iterative design process that focused on bridging the gap between a developer and a designer. While I, as a designer, made design decisions through every week, my fellow colleague, Yuhang, a developer gave his contribution by implementing K-means to extract hues from different paintings and tuned parameters based on the requirements to improve the model interpretability. Finally, we visualized the results using Plotly to generate multiple kinds of treemaps and scatterplots. These were then displayed through an interactive platform using Adobe XD that allows users to explore and compare 4 different artists through 4 years, a total of 16 unique paintings!

Tools Used: Python, Plotly, Procreate, Adobe XD