Text only

CM30076 / CM30082
Individual Project

Project Ideas

Prof. Phil Willis

pjw@cs.bath.ac.uk

Point modelling/rendering solid objects
Point modelling is typically used to laser-scan real objects: the result is a high-density set of coordinates, representing points on the surface of the object. This project takes the idea a step further: what if we represent all solid space by a regular, dense cloud of points? We can create simple geometric objects such as a sphere by including only the interior points, for example. We can create complex objects by boolean operations on the points of simpler ones. For example, we can cut a cylindrical hole from the same sphere by eliminating those points which fall within the cylinder. We can even carve away material freehand. There are two projects here: one on the modelling side and one on the rendering side.
Rendering point models of mists and clouds
We can represent mists and clouds by high density point models. Each "point" scatters light and corresponds to a small drop of water suspended in the air. If the mist is illuminated, perhaps by one or more coloured lights, light will scatter through the volume in visually interesting ways. The object of this project is to simulate this and show the results as pictures. Mists are not always uniform so the points can be distributed in patterns or statistically. Lighting can be highly directional (e.g. a laser beam) or broad (a conventional light). Scattering can be made frequency dependent and does not have to obey "correct" physics. There is thus lots of scope here to experiment, including with the kinds of effects needed in movies or games, or to simulate real clouds, or to experiment with imaginary kinds of mists that produce unusual effects.
Prerequisite knowledge: Programming skills. If you don't know how to create and store pictures, I can supply a simply C library; otherwise you are free to choose any language you like.
Compositing colour images and volumes
Digtal compositing is the layering together of two or more images to produce another image showing elements from both source images. It is much used for special effects in movies and television. We have developed some new techniques which extend the range of visual effects which can be achieve and also permit volume objects (for example, data captured by a medical scanner) to be treated as image layers in the same way. This project will involve building a compositor, preferably with a simple GUI to permit complex composites to be designed. Background reading material of three papers and a good general book on compositing will be provided. The implementation can be in any language that you are familiar with but the GUI probably requires a Microsoft-based implementation.