- What is Noise?
- Signal to Noise Ratio
- Color Banding
I think television has betrayed the meaning of democratic speech, adding visual chaos to the confusion of voices. What role does silence have in all this noise? – Federico Fellini
Noise is unwanted information in a signal. A fly in your soup is noise. Too much salt in popcorn is noise. To some this noise might be okay. For some, not enough!
Every visual system converts light into another form of energy, usually electrical energy. This conversion introduces additional data. As Professor Sampler might have told you, no two things are alike. The difference could be noise.
We have already seen that sometimes noise can increase the perception of acutance. Some would say this is a desirable thing, while the purist will not accept any kind of noise, no matter what the consequences.
The ratio of the ‘good’ signal to noise is called the signal-to-noise ratio. A good experience can be had with a signal and some noise. However, we can hopefully agree that as the signal data reduces, the experience is worsened. One can gather from this that nobody is complaining about ‘too much signal’.
Take a look at these images:
The blue ball in the center has a white highlight. The tonal range goes from white to blue to dark blue in many shades. The blue ball on the left tries to reproduce the same gradations, but without the luxury of having as many colors (tones) at its disposal. It is a way of representing tones or colors without that color being actually available.
How is that possible? Look at the third image.
With just two colors, red and blue, one can produce the illusion of purple, without it being actually used! Remember, I’m not talking about mixing blue and red to get purple. This idea has been used by artists and painters over the ages to great effect.
So, how does this property help improving signal to noise ratio?
Look at the first image. This is the original photo. When we convert the image to web-safe colors, we drastically reduce the number of tones available, just like our blue balls above. A lot of detail is lost.
In the third image, we add a bit of magic: dithering. Suddenly the large expanses of black and grey vanish, and the image looks more like the original photo. How was this possible? By adding more ‘dots’ to the image, and spacing them apart differently, the impression of a greater tonal range is created.
Impressive, right? Noise used to create the impression of detail.
This is what happens when the color bit depth is changed. An image originally produced in 16-bits per channel, will give each color 65,536 tones from white to most saturated. When that image is converted to an 8-bit image, there are only 256 values now available to represent the same tonal range. Sometimes, this produces results like this:
This is called Color Banding, or sometimes, just banding. It usually happens in large expanses of the same color, like skies, clothes, skin, etc. Dithering is one of the solutions to hide the effects of banding.
- Noise is unwanted information in a signal that sometimes has useful purposes.
- The ratio of the pure signal to noise is called the signal-to-noise ratio. It is one measurement of the overall ‘quality’ of a signal.
- Dithering is the method of adding noise to improve the perception of tonality, color and detail.
- Color banding is the effect witnessed when there aren’t enough tones to cover the entire range necessary to display an image.
Links for further study: