Students working with Dr. Michael Jones in the Computer Generated Natural Phenomena (CGNP) Laboratory were able to take their education out of the classroom and into Utah's Castle Country on a field trip to Goblin Valley. The day-long excursion helped the participating students catch the vision of what they are trying to recreate in their lab.
Back in Provo, the students are investigating algorithms and data structures that simplify the creation of natural phenomena for use in computer generated animation. Natural phenomena can be powerful visual cues for place and mood in computer generated (CG) animation. Unfortunately, creating and directing natural phenomena in CG animation is difficult and time-consuming.
In computer generated animation, landscape is often a critical indicator of setting and mood; however, the process of creating and directing natural phenomena in CG animation is difficult and time-consuming. The CGNP Lab investigates algorithms and data structures that simplify the use of natural phenomena in CG animation. Their goal is to give a director the tools needed to interactively create and direct natural phenomena with minimal intervention by technical support staff. Their collaboration with the BYU Animation Program allows for the development of their ideas within the context of a full CG production pipeline.
A current project, the BYU Animation Program's film, "Kites," relies heavily on directable clouds which can be used as both background landscapes and props. As the three photographs (left) of clouds on Mt. Timpanogos show, the shape, color, and appearance of clouds strongly communicate mood and feeling, as they do in the film "Kites." The CGNP lab is creating cloud models and shaders which strike a balance between automated physical realism and artistic directability.
Previous work on computer generated landscapes focused on fractal models which, although they are excellent models of certain kinds of mountainous terrain, are limited in the types of natural phenomena they can simulate. A wide range of more realistic landscapes can be generated by including stratiagraphy and differential erosion in landscape generation. Terrain features ranging from mountains with cliff bands to sandstone arches to slot canyons can be generated with differential erosion. The primary technical challenge in the CGNP Lab's work is modeling rock strata so the differential erosion from a variety of sources can be simulated. Dr. Jones and his students have discovered that a wider range of more realistic landscapes can be generated by simulating the natural weather process using grids of axis-aligned voxels (3D versions of pixels), as in the computer generated recreation of a Goblin Valley formation pictured right.