Lecture 3 - colour models
Monday, October 8, 2007 at 18:24
martian77 in Multimedia design

We didn't quite reach the end of lecture 2 in lecture 2, so I've got a bit to finish up on!

The HSV (hue, saturation and value) colour model (or hue saturation and brightness in Photoshop). 

We started looking at this last lecture, but we got a bit more detail on this one. 

Hue - based on human perception of the wavelength of the dominant colour. Psychophysical was the word used in the lecture, which apparently means involving the action or mutual relations of the physical and the psychical (in this case the stimulation of the cones by the light and the way it appears to us).

Saturation (or excitement purity) - a measure of the amount of white light. High saturation means bright colours and low white light content. Paul had a graphic of a graph with the visible light wavelengths on the x axis and the energy on the y axis. A highly saturated colour has a high level of energy at a given wavelength and low energy levels across all of the other wavelengths. As saturation decreases, the energy levels for all of the other wavelengths increase.

Brightness (value) - the perceived intensity of the light. This is very subjective. For example, the screen in the lecture theatre looked quite bright with the lights on, but when he dipped the lights the screen appeared to be brighter. In fact it had less light shining off it so it was less bright, but it appeared brighter in contrast to the rest of the room.


Why have different colour models? RGB seems to represent the light pretty well.

One reason is to simplify certain transformations. For example, to make a picture brighter in RGB would require all of the colour values to be changed individually. By representing in HSV we can change a single value across the entire picture.

Another reason is that it allows the data to be reduced to only 2 colour variables - hue and saturation. 

Apparently something called the YUV colour model is used in PAL TV systems, and YCBCR is used in digital televisions and jpegs. I didn't fully get this, so I'll do a little more reading and write something a bit more informative about it! But they are systems based on how humans see light, and separate out the luminance/chrominance bits. Humans see colour and intensity differently (back to the rods and cones). This means we can produce sharp colour images by overlaying a sharp black and white image with a very fuzzy colour component, and allows us to downsample the colour information.

YUV 4:2:2 is widely used, with the numbers reflecting the ratio of the components. If you take 4 pixels, each one would have a YUV component. Instead, we can send two of those pixels with YU information, and two with YV information. This means the information is stored in 8 bytes rather than 12.

Article originally appeared on Life on Mars (http://www.martiandaze.net/).
See website for complete article licensing information.