cam/

README

cam: Color Appearance Models

Color appearance models (CAM) provide a way of organizing colors, with the overall goal of better fitting the way that people actually perceive color. The basic RGB color model is useful for controlling the pixels on a color monitor, but it does not define a useful metric for generating colors relative to each other in ways that match our perceptions. For example, making each of the RGB components 10% lower does not consistently produce the same perceptual effect of darkening the color by 10%.

The widely-used HSL (hue, saturation, lightness) or HSV (hue, saturation, value) systems are basic examples of CAMs that do a better job than RGB, but they are designed more for computational simplicity and efficiency than perceptual accuracy.

Our color perception starts with the cones in the foveal region of the retina, which loosely correspond to the R,G,B color components, but in fact the Red and Green (Long and Medium wavelength) cones overlap considerably in their responses, while Blue (Short) cones are more separated in their response to different light frequencies. Thus, one first pass step in a CAM is to map the RGB components into something that more closely matches the LMS (long, medium, short) cone responses.

At higher levels in the visual processing pathway (in the visual cortex), the basic cone responses from the retina get organized into color opponent responses, reflecting the difference between the amount of Red vs. Green and Blue vs. Yellow (where Yellow is roughly an average of RG). In color speak Red = Long, Green = Medium, Blue = Short wavelengths, corresponding to the color absorption profile of cones in the fovea of the retina.

The current best CAM in terms of the ability to predict human perceptual judgments is CAM16, which is implemented in the cam16 package. This is used in the latest version of material design to provide algorithmically-generated color schemes with good overall psychophysical properties.

cie: International Commission on Illumination

See cie for standard color spaces and values defined by the International Commission on Illumination (CIE), including XYZ and LAB (L*a*b*) which can be easily computed from corresponding standard RGB (sRGB) values.

RGB as displayed on a computer monitor is typically "gamma corrected" to compensate for the luminance properties of the display, so this gamma correction must be removed before using RGB values to covert into other CIE spaces.

• Here’s what is implemented:

• paper for this: Moroney et al., 2002

• Color spaces based on LMS cones: https://en.wikipedia.org/wiki/LMS_color_space

• Standard CIE 1931 XYZ color space:

• Standard color checker: https://en.wikipedia.org/wiki/ColorChecker

CAM02, CAM16

Robert W.G. Hunt established many of the key principles of color appearance models, as attested by the many references in Helwig & Fairchild (2022) (HF22), which is a particularly good reference for actually explaining things in plain English. Moroney, N., Fairchild, M. D., Hunt, R. W. G., Li, C., Luo, M. R., & Newman, T. (2002) established the CIE CAM02 reference model, which set the standard for many years, until being revised in the CAM16 model. Here's the HF22 explanation of the key factors in CAM02 and CAM16:

Brightness is the perceptual attribute by which a light source or reflective surface appears to emit or reflect more or less light. Lightness is the brightness of a stimulus relative to the brightness of a white-appearing stimulus in a similarly illuminated area, also known as the reference white. While the brightness of stimuli has a general positive correlation with the amount of light they emit or reflect, there is no simple relationship between the amount of light emitted by a stimulus and its brightness and lightness. For instance, stimuli with greater purity appear brighter than stimuli with less purity if they are of the same luminance (known as the Helmholtz–Kohlrausch Effect).

The perceptual attribute colorfulness describes the absolute chromatic intensity of a visual stimulus. Chroma and saturation are relative measures of colorfulness; chroma is defined as the colorfulness of a stimulus relative to the brightness of similarly illuminated white and saturation is defined as the colorfulness of a stimulus relative to its own brightness.

CAM02

See cam02 for implementation, and Moroney et al, 2002 for description, along with this wikipedia page.

We implement the functions that transform RGB or XYZ directly into LMS and color opponents.

CAM16

See cam16 for implementation, and Li, Li, Wang et al, 2017 for description.

CIELAB

L*a*b* is defined in the CIELAB color space: https://en.wikipedia.org/wiki/CIELAB_color_space

HCT

See hct for implementation.

material design uses components of CAM16 and LAB to create an HCT (hue, chroma, tone) space that works well for automatically creating different color shades for GUIs. The implementation in a variety of languages (excluding Go) is on github at: https://github.com/material-foundation/material-color-utilities -- we leveraged this code for our HCT implementation in Go.

• Hue is CAM16.Hue
• Chroma is CAM16.Chroma
• Tone is LAB Lstar (L*)

HCT Colorspace

Color in V1

Two effective populations of cells in V1: double-opponent and single-opponent

• Double-opponent are most common, and define an edge in color space (e.g., R-G edge) by having offset opposing lobes of a gabor (e.g., one lobe is R+G- and the other lobe is G+R-) – this gives the usual zero response for uniform illumination, but a nice contrast response. We should probably turn on color responses in general in our V1 pathway, esp if it is just RG and BY instead of all those other guys. Can also have the color just be summarized in the PI polarity independent pathway.

• Single-opponent which are similar-sized gaussians with opponent R-G and B-Y tuning. These are much fewer, and more concentrated in these CO-blob regions, that go to the “thin” V2 stripes. But the divisions are not perfect..

References

All references can be obtained via the CCNLab group on https://zotero.org if you can't find them elsewhere -- just ask to join.

• Gegenfurtner, K. R. (2003). Cortical mechanisms of colour vision. Nature Reviews Neuroscience, 4(7), 563–572. https://doi.org/10.1038/nrn1138. (highly recommended scientific overview)

• Hellwig, L., & Fairchild, M. D. (2022). Brightness, lightness, colorfulness, and chroma in CIECAM02 and CAM16. Color Research & Application, 47(5), 1083–1095. https://doi.org/10.1002/col.22792 (accessible overview of CAM16 terminology)

• Li, C., Li, Z., Wang, Z., Xu, Y., Luo, M. R., Cui, G., Melgosa, M., Brill, M. H., & Pointer, M. (2017). Comprehensive color solutions: CAM16, CAT16, and CAM16-UCS. Color Research & Application, 42(6), 703–718. https://doi.org/10.1002/col.22131

• Moroney, N., Fairchild, M. D., Hunt, R. W. G., Li, C., Luo, M. R., & Newman, T. (2002). The CIECAM02 Color Appearance Model. Color and Imaging Conference, 2002(1), 23–27. (definition of CAM02)