Visualization Color Repair: A Browser Extension for Colorblind Users
Presentation
- Session
- Data, Meet Human: Vis That Cares
- Time
- Wednesday, Nov 11, 10:09 – 10:18 (US/Eastern) · session 10:00 – 11:30
- Location
- Hall America north
Keywords
Visualization, color, accessibility
Abstract
For people with colorblindness, the color palettes used in data visualizations can be illegible such that encoded data signals collapse. Although extensive prior research studies the perceptual mechanisms behind colorblindness and develops tools to help visualization designers choose colorblind-safe palettes, relatively little work empowers people with colorblindness to make found visualizations more legible for themselves. To explore the design of such assistive technologies, we contribute Visualization Color Repair (VCR), a browser plugin built to selectively recolor data-encoding elements in SVGs to avoid signal collapse under simulated colorblindness. We ground this signal preservation approach in a formative analysis of visualization examples, deriving perceptual invariants used for color linting in VCR and assembling a corpus of test cases to support validation. Our investigation highlights how signal-preserving palette repairs exhibit a ``curb cut effect’’ but also come into tension with original design choices about color semantics, surfacing value conflicts around color accessibility.
For Practitioners
Our work presents a practical web browser extension for recoloring SVG visualizations found “in the wild” in order to correct for color choices that will render the data illegible for colorblind users. For data communicators who make public-facing data visualizations on the internet, our paper offers conceptual insights about what to prioritize when choosing accessible colors, and our tool makes this guidance concrete and actionable.