Bilinear filtering is one method in computer graphics for making pixels look smoother when drawing to the screen. (Another is bicubic filtering, which looks much nicer, and will be discussed later).

So to kick this off, a bilinear graph looks like this:

      / | \
     /  |  \
    /   |   \
   /    |    \
  /     |     \
 /      |      \

Now take the origin as the pixel whose color is to be calculated. The theory goes is that each pixel directly next to that one factors in heavily, and, up to a certain threshold, each pixel after that factors in inversly proportional to its distance from the pixel in question. Obviously the pixels original color factors in the greatest when determining the final color. Most filtering algorithms have to make a second pass at the entire pixel table to do this sort of filtering, although some really great implementations use some a lot of numbers and some scary math to accomplish it in one pass.

Bilinear filtering is an additive filter applied to the pixel grid, while Bicubic is subtractive. The only problem is that this method of graphical filtering tends to look a little fuzzy when all is said and done. Because of this woe, the slightly more complicated bicubic filtering was developed, whereby pixels at the end sharpen picture by subtracting from the final color.