Some info about the technique

What is "Affective Scene Generation"?

Affective Scene Generation is a technique for generating virtual scenes and environments by using adjectives to describe the scene. In essence, the user will specify a number of adjectives and weights for these adjectives to indicate how applicable each adjective is. These adjectives and weights are then transformed into parameters which are used by procedural modelling techniques to generate the scene.

Why is this important?

As graphics hardware becomes more powerful, personal computers are able to process ever larger and more complex scenes and environments. Consequently, applications that make use of virtual environments (such as games and movies) require larger and more complex environments in order to appear more realistic. There is therefore pressure on designers to generate these environments, but "human hardware" is unable to keep up: our brains can change, but only through a slow evolutionary process and not over the course of a year like computer hardware. As such, designers are put under too much pressure to produce acceptable results.

But procedural methods are old: why pursue this?

Procedural methods have been researched extensively over the past number of years, to the point where we can model many natural phenomena through procedural methods. However, the parameters used by these techniques are often difficult to use (that is, it is not intuitive how changes to parameters affect the final output), and the parameters can also often interact in complex ways. Simply directly manipulating the parameters is not feasible: a simpler yet more powerful method is needed.

This is where adjectives become important: people are very good at describing experiences and objects using adjectives and emotions. We can much more easily relate an object to a set of adjectives than to a set of numbers. A technique like this will also allow uneducated people to generate visually complex and stunning scenes without needing to know any of the underlying mathematical and computational theory, and could thus be a useful educational tool.