This NSF project seeks to accurately model the visual appearance of objects such as the light glints on a shiny surface like polished metal, a car body or a laptop screen; and the appearance of natural materials like hair and fur.  All of these topics require a deep understanding of the way light interacts with and reflects from surfaces, since what we see is the reflected light from an object.  Many previous approaches have been concerned only with light interaction at the surface of an object, but realistic appearance involves modeling light scattering within the body or volume of the material, for example the scattering of light within individual fur or hair fibers and between them.  Finally, along with understanding theinterplay of light and surface appearance, we must develop ways to measure the reflectance or appearance properties of objects, since this is what makes metal look like metal, skin look like skin, or paper look like paper.  The overall goal of our project is to create realistic computer-generated images of a variety of real-world materials.  This has enormous applications in product design and visualization, wherein computer models can be used to predict the appearance of car bodies, laptop screens, and metal components.  It is also significant for entertainment applications, such as computer-generated movies with furry animals.  

Our project has had three main thrusts.  First, we have developed a method to improve how computer graphics software reproduces the way light interacts with extremely small details, called glints, on the surface of a wide range of materials, including metallic car paints,metal finishes for electronics and injection-molded plastic finishes.The method requires minimal computational resources and can be used in animations. Current methods can only reproduce these so-called glints in stills.  The standard approach to modeling the way surfaces reflect light assumes that the surfaces are smooth at the pixel level. But that's not the case in the real world for metallic materials as well as fabrics, wood finishes and wood grain, among others. As a result, with current methods, these surfaces will appear noisy, grainy or glittery.  There is currently no algorithm that can efficiently render the rough appearance of real specular surfaces, which is highly unusual in modern computer graphics, where almost any other scene can be rendered given enough computing power.  We break each point into a number of microfacets, and show that one can efficiently and mathematically compute the reflected light.  The work has attracted wide press coverage (many parts of this paragraph are quoted from the UCSD press release) and it is likely that variants will be adopted soon for product design and visualization renderings, as well as in applications like computer-generated imagery for movies and video games.

A second thrust has been on the development of models based on physics to create the appearance of animal fur.  This requires careful measurement and modeling of individual fur fibers, which are really made of two concentric cylinders with an inner scattering medulla.Ours is the first model which predicts the way in which light interacts with a single fur fiber, including scattering in the medulla volume, and validates the results with actual measurements.  In subsequent work, we have simplified our model, and made it faster to compute light interaction between fur fibers.  This enables us to achieve highly realistic images of animal fur, enabling more realistic animated charaters in entertainment applications, and predictive modeling of the appearance of different types of fur.  This work is already being adopted for production rendering of computer-generated imagery.

 
Finally, accurate modeling of the appearance of real-world materials requires not just an understanding of light interaction with surfaces, but a measurement of the reflectance properties of objects.Traditionally, this has required many thousands to millions ofmeasurements on expensive devices known as gonioreflectometers.  We have shown that accurate results can be achieved with only 5-20 carefully designed measurements, and in the limit with only two photographs.  

In summary, a careful combination of new reflectance measurement methods, coupled with new computational techniques and understanding of the physics of light reflection in surfaces like fur, enables new levels of realism in computer-generated imagery for product design and visualization, entertainment and gaming, and industrial measurements of appearance.

Last Modified: 09/09/2017
Modified by: Ravi Ramamoorthi