Visualization Analysis and Design: Color

Introduction to Color in Visualization

• Shift focus from spatial arrangements to visual encoding through color.
• Importance of decomposing color beyond treating it monolithically.

Decomposing Color into Channels

• Three Channels of Color:
  1. Luminance Channel:
     • Represents brightness (grayscale from black to white).
  2. Saturation Channel:
     • Represents colorfulness (e.g., the amount of pink between gray and bright pink).
  3. Identity Channel:
     • Represents hue, which helps in identifying categories (e.g., red, blue, green).
• Different properties of these channels affect perceptual understanding.

Categorical vs. Ordered Color

• Color can be perceived as categorical or ordered, depending on context.
  • Example of years distinguished by hue (categorical) vs. shades of green representing years (ordered).
• Importance of Distinguishable Bins:
  • Human perception relies on relative comparisons, especially for contiguous colors.
  • Maximum of 6 to 12 distinguishable bins for categorical colors that can be perceived effectively.

Challenges in Using Color for Visualization

• Common Mistakes:
  • Trying to encode more categorical levels than the number of available discriminable bins, particularly in separated regions.
  • Example of color-coding chromosomes leading to confusion in non-contiguous small regions.
• Ordered Color Limitations:
  • Fewer distinguishable bins than expected, demonstrated through examples of U.S. states.

Issues with Rainbow Color Maps

• Rainbow colors often lack intrinsic ordering; individuals may not perceive them consistently.
• Non-linearity in color perception complicates rainbow gradients, as seen in examples.
• General Principle:
  • Use perceptually uniform color maps (e.g., Viridis, Magma) designed for sequential data.

Using Saturation and Luminance

• Saturation and luminance cannot be used simultaneously for effective encoding.
• Transparency complicates perception and should be carefully handled in visual design.
• Key Principle:
  • Typically limit to two bins for separated regions and choose one channel between saturation, luminance, and transparency.

Color Palettes in Visualization

• Single Attribute Palettes:
  • Aim for maximum distinguishability.
  • Types of Palettes:
    • Sequential:
      • From minimum to maximum values.
    • Diverging:
      • Emphasizing a neutral midpoint with two saturated colors.
    • Cyclic:
      • Highlighting cyclic nature with multiple hues.

Bivariate Color Maps

• Illustrate two attributes using color.
• Bivariate color maps with one binary attribute are easier to interpret.
• Trickiness of Multiple Levels:
  • More complex to design and interpret when both attributes have multiple levels.

Conclusion

• Effective color use in visualization requires careful consideration of data characteristics and perceptual principles.