Teaching Color Science to EECS Students Using Interactive Tutorials: Tools and Lessons

Yuhao Zhu

TL;DR

This paper introduces a series of six interactive, web-based tutorials designed to bridge the gap between oversimplified "color cube" concepts and the mathematically rigorous reality of color science. By combining 3D visualizations with dynamic linear algebra, the tools help EECS students move from treating color as a mere sensation to understanding it as a physical and psychophysical system derived from first principles.

What Problem Does This Paper Address?

In many Electrical Engineering and Computer Science (EECS) curricula, color science is a "one-lecture" afterthought, often reduced to the sRGB color cube. This leads to two major issues:

  1. The Dimensionality Gap: Color is fundamentally a 3D concept, but classic textbooks rely on static 2D projections. Students struggle to visualize how color spaces actually "sit" in 3D or why certain colors are "imaginary."
  2. The Rigor Gap: Existing tutorials often prioritize intuition over math, leaving students confused when they need to apply color concepts to real-world engineering tasks like camera sensor calibration or display power optimization.

The author, Yuhao Zhu, argues that without a solid foundation in the physics of light (Spectral Power Distributions) and the biology of the eye (Cone Fundamentals), students cannot truly master advanced topics in graphics, AR/VR, or computer vision.

Methods

The author developed a suite of six interactive tutorials (T1–T6) using modern web technologies like Plotly.js and Chart.js. The tutorials follow a specific pedagogical structure:

  • Narrative-Driven Steps: Each tutorial is a guided derivation.
  • Interactive Actions: Students perform tasks, such as clicking on a spectrum to pick "primary" lights or drawing a custom white point.
  • Dynamic Math: As students interact with the plots, the underlying mathematical equations (e.g., matrix coefficients in a linear system) update in real-time on the page.
  • 3D Visualization: The tools allow students to rotate and examine color volumes, such as the human visual system (HVS) gamut, to see how it relates to the spectral locus.

The tutorials are divided into Basics (Cone fundamentals, Gamut/Imaginary colors, XYZ space, and Color cubes) and Applications (Camera color correction and Color blindness simulation).

Datasets

While not a traditional machine learning paper, the tutorials utilize several standard color science datasets and models:

  • Cone Sensitivities: The Stockman & Sharpe (2000) 2-degree cone sensitivity data.
  • Standard Illuminants: CIE Standard Illuminants (like D65) for reference white points.
  • Camera Data: Measured spectral sensitivities from 48 modern cameras (smartphones, DSLRs, and industrial cameras).
  • Calibration Targets: Data from the standard 24-patch ColorChecker.

Key Insights and Ideas

  • Trichromacy as Linear Algebra: The tutorials frame the fact that humans have three cone types as a system of linear equations. Students see that if we had more or fewer primaries, the system would be over- or under-determined, providing a mathematical "why" for trichromatic vision.
  • The Reality of "Imaginary" Colors: By visualizing the 3D LMS space, students can see that the "primaries" of the HVS actually sit outside the region of real, physical light. This makes the abstract concept of "imaginary colors" (like those in the XYZ space) geometrically intuitive.
  • Color Blindness as a Projection: Instead of just showing "what a colorblind person sees," the tutorial demonstrates that dichromacy is a geometric projection from 3D color space onto a 2D plane.
  • No "Magic" Constants: By allowing students to derive the Color Matching Functions (CMFs) themselves by picking primaries, the tutorials demystify the "magic" numbers often found in graphics APIs.

Key Results

The tutorials were integrated into a 3-week course module for undergraduate and graduate students.

  • Improved Comprehension: In a survey of ~30 students, over two-thirds reported that the tutorials significantly deepened their understanding.
  • Concept Mastery: In a pre- and post-tutorial survey regarding camera color correction (T5), fewer than 10 students could explain the principle initially. After the tutorial, "almost all" students provided the correct answer.
  • Student Contribution: The approach was successful enough that a former student actually helped design the color blindness tutorial (T6), and others used the tutorial math as a foundation for their final course projects.

Limitations and Future Work

  • Sample Size: The evaluation was based on an informal survey of a relatively small cohort (30 students).
  • Scope: The current series focuses heavily on the "standard observer" and linear color science; it does not yet delve into complex non-linear phenomena or advanced color appearance models (CAMs).
  • Open Source: The author plans to release the source code to the community to allow for the development of more tutorials and broader adoption in EECS education.

The tutorials are publicly available at: https://horizon-lab.org/colorvis/

Summary updated

Abstract

Teaching color science to Electrical Engineering and Computer Science (EECS) students is critical to preparing them for advanced topics such as graphics, visualization, imaging, Augmented/Virtual Reality. Color historically receive little attention in EECS curriculum; students find it difficult to grasp basic concepts. This is because today's pedagogical approaches are nonintuitive and lack rigor for teaching color science. We develop a set of interactive tutorials that teach color science to EECS students. Each tutorial is backed up by a mathematically rigorous narrative, but is presented in a form that invites students to participate in developing each concept on their own through visualization tools. This paper describes the tutorial series we developed and discusses the design decisions we made.
BibTeX
@inproceedings{zhu2023teaching,
  author = {Yuhao Zhu},
  title = {{Teaching Color Science to EECS Students Using Interactive Tutorials: Tools and Lessons}},
  booktitle = {Visualization and Data Analysis},
  year = {2023},
  doi = {10.2352/ei.2023.35.1.vda-396},
}
Figure