.png){kind=link}
 - Vertical.png){kind=link}
.png){kind=link}
 - Vertical.png){kind=link}
.svg){kind=link}
.png){kind=link}
Colour Appearance Models
A Colour appearance model (CAM) is a mathematical model that seeks to describe the perceptual aspects of human colour vision, i.e. viewing conditions under which the appearance of a colour does not tally with the corresponding physical measurement of the stimulus source. They are more sophisticated than colour spaces.
This is different from Colour Models which describe colours as tuples in a coordinate space.
Phenomena to Capture
- Chromatic adaptation
- It is the ability of humans to discount a White Point (or Colour Temperature) of an illuminating light source when observing a reflecting object. For example, a white paper looks white no matter whether the illumination is blueish or yellowish.
- This is one of the most basic and most important aspect of all colour appearance phenomena
- So a chromatic adaptation transform (CAT) that is used to model this behaviour is central to any CAM.
- Hue Appearance
- Contrast appearance
- Colourfulness appearance
- Brightness appearance
- Spatial Phenomena or Optical Illusions
- They only affect certain areas of an image, and this has to do with the way the human brain interprets information contextually instead of as raw colours
- They are hard to model because of their contextuality, and CAMs that try to do this are called Image Colour Appearance Models (iCAM).
Implementations
- CIELAB (Technically a colour space, but it was the first approach to building a CAM)
- It uses the wrong von Kries transform 1 by performing it on the XYZ colour space directly instead of changing to the LMS Colour Space first for more precise results
- Nayatani et al. model
- Hunt model
- RLAB
- Unlike CIELAB, it uses a proper von Kries transform 1
- LLAB
- CIECAM97s (was popular until CIECAM02 arrived)
- IPT (It is well suited for gamut mapping)
- ICtCp (Improves on IPT)
- CIECAM02
- iCAM06 (It is an iCAM)
- CAM16 (It also comes with the CAM16-UCS space; not a CIE standard although published by CIE)
- CIECAM16 (The 2022 CIE standard release of CAM16)
- OKLab (Same structure as CIELAB, but fitted with improved data such as CAM16 output for lightness and IPT data for hue; It was meant to be easy to implement and use like CIELAB and IPT where, unlike sRGB)
- As of December 2022, it is part of the CSS colours level 4 draft.