CSC317 - Computer Graphics Rikin's Notes

Raster Images

Raster Images

Images

Raster images

An image is a distribution of light on a 2D canvas.

A raster image captures an image in a grid, i.e. with pixels.

Raster devices

  • input
    • 2d array sensors
    • e.g. scanners, cameras
    • use bayer filters to convert input light into pixels
  • output
    • printers
    • displays can be
      • emissive, emitting light by themselves (e.g. LED)
      • transmissive, allowing only certain light to be transmitted (e.g. LCD)
    • display images as arrays of pixels

Other representations

Another common representation of images is vectors, i.e. a mathematical description of the image rather than a grid of pixels.

Images as functions

An image can be abstractly thought of as a function $I : \R^2 \to \R^n$, where each entry of the output vector specifies some quantity. For example, a typical picture is a function from $\R^2$ to $\R^3$, where $I(x, y) = (r, g, b)$ specifying red, green, blue pixel values.

Raster image representation

Colour specification

An RGB colour can be specified in multiple ways:

  • straightforward RGB values, using red, green, blue, the primary colours of light
  • CMYK which uses cyan, magenta, yellow, the complementary colours
  • HSV - hue, saturation, value, which is much more intuitive for humans
  • HSL - similar to HSV

Discretization

Continuous maps cannot be represented as pixels, so discretization is needed when we want to represent something that is continuous in reality.

Discretization of colours can leave artifacts such as banding

01 banding.png

Gamma correction

On all devices, display intensity is not linear with respect to input intensity

Transparency

Transparency is represented by a fourth coordinate in each pixel, i.e. RGB becomes RGBA

The last channel is the alpha channel, which specifies how visible the pixel is.

Once transparency is defined, we can define compositing and blending.

[…]