The aim of this project is to create a depth of field algorithm that can operate in real time and produce the true depth of field effect. The hope is to implement the algorithm as a post process effect using the current Rendering or Parallel Processing application programming languages (APIs). Possible APIs to be used in the solution are discussed as well as some research into the problem area itself. Demonstrating an implementation using OpenGL and GLSL that shows some promise as it achieves thirty frames per second.

The basicconcept behind this implementation of Depth of Field is sound. However, Thereare problems. blurring of the image planes was achieved though OpenGL's mipmapgeneration, which has proved unsuitable. Blurred images generated are blockyand do not generate a good effect. Another problem caused by this method of blurringare the dark band artefacts. The image is split into multiple buffers based on theircircle of confusion value at each pixel, the band artefacts are generated wherethe edge of two buffers are blurred with the clear colour which is black. Abetter per-pixel approach to blurring these buffers is required to complete theeffect.
In terms ofoverall speed and proof of concept as a hole the projects has been a success. Thispost process method of creating depth of field images does achieve real-time speedsand with a better blurring method could produce high quality effect.