Getting There: The Ten Top Problems Left

irst, some caveats. My view of computer graphics is broad and goes beyond what some may consider as the realm of computer graphics. The problems I list here are unranked, with their order reflecting neither how much has already been done to solve the problem, nor the importance of the problem, nor the potential difficulties in solving the problem. In each of the areas I mention, there is ongoing research, but space limitations don't allow a detailed discussion of what remains to be done, much less what has been done, or by whom. But, rest assured that those tackling these broad challenges will be able to stand on the shoulders of others and will have small hills rather than mountains to climb. Also note that this is one of several " top ten " lists, including those from Sutherland, 1 Heckbert, 2 Crow and Feiner's 1991 Siggraph panel, 3 and Blinn. 4 The lists reflect both our progress since 1966 and the authors' varying views of the discipline. Here's mine. 1. Fill the gap between image-based and geometric modeling techniques. Consider an arbitrary real-world scene, perhaps your office, the view out the window, or your workshop. Could our current technologies create a photorealistic rendering of that scene? Only if there are objects whose shapes can be modeled using our current repertoire of geometrical modeling techniques (such as polygon meshes, NURBS, conics, quadrics, superquadrics, particle systems, L-systems, weathering systems, fur/hair systems, and so forth) and whose surface properties can be characterized or simulated with known techniques (such as texture mapping, bump mapping, weathering, anisotropic reflection, and so on). More likely, it would be necessary to use image-based techniques to capture photographically the fine texture detail and possibly the geometry of objects in the scene. Image-based techniques create impressively beautiful and stunning images because they capture detail that we don't yet know how to model. The techniques are exceedingly useful and have required considerable research effort to develop. But, we need to keep working toward modeling all aspects of the real-world environment rather than limiting ourselves to the record, process, and playback of image-based rendering (IBR). Ultimately, we should be able to completely model any real-world environment by synthesis rather than by photography and reconstruction. And, given a model, we can generate instances of phenomena that look realistic even though the specific instance does not exist. Of course, IBR techniques might be …