Proceedings of the 29th annual conference on Computer graphics and interactive techniques

Abstract

Welcome to the SIGGRAPH 2002 Papers Program. It's an exciting collection that pushes the boundaries of our field. Several areas have an increased presence: the numerical simulation of natural phenomena; data- driven synthesis, including texture synthesis from images and motion synthesis from motion-capture data; and image- and video-processing methods.There were 358 submissions and 67 acceptances. About six dual submissions to other conferences were discovered and rejected automatically. The 358 submissions represent a 20% increase over last year.The papers are arranged in their order of presentation, grouped into sessions. Sessions are created after the paper selection process and thus reflect the result of the selection. We take all of the papers that are good enough and then figure out sessions for them. I once heard of a conference in which reviewers give one of three ratings: "Accept," "Reject," and "Accept as Filler." By avoiding the preselection of topics, i hope we can forever avoid that third category.There are several notable changes this year. First and foremost, as the start of a two-year experiment, this year's papers are published as an issue of ACM Transactions on Graphics (TOG). Because TOG is a journal, papers had to be refereed rather than just reviewed. Each submission was either accepted, accepted with minor changes, or rejected. Kurt Akeley's idea, for S1GGRAPH 2000, of only accepting papers "as is" was a great step towards fairness. But I felt that it also let some bad things happen -- papers were rejected because of just a few misplaced words. Letting reviewers require minor changes addresses this issue. The authors made these changes and a referee checked them. If the author felt that the minor changes were not acceptable, s/be was welcome to withdraw the paper. The advantages to this new approach are threefold: papers are double-checked for mistakes; small mistakes or omissions can be fixed; and papers, having been refereed, can be published in a journal, which is important in some academic tenure cases.There are disadvantages too. The refereeing process was rushed and sometimes inconvenient, and different reviewers had different notions of "small changes," despite my best efforts to outline what these were. In retrospect, Kurt's rule of "acceptable as is" has the enormous advantage of being unambiguous? Jessica Hodgins, next year's chair, will have to evaluate this experiment carefully.A second change was in the selection process: the non-committee reviewers were selected by both committee members responsible for a paper instead of just one as in earlier years.A third and final change was that by expanding the technical program to start one day earlier, we have no overlapping sessions. When Jim Kajiya shifted to overlapping sessions, he noted that it was a positive change: no longer was the committee "counting papers" near the end of the selection process and getting tougher about accepting them because "there isn't room." I asked my committee not to count, but rather to work with the "accept all the papers that you think should be SIGGRAPH papers rule" and see how it turned out. Even with the extra day, not everything fit, so we lengthened some sessions to accommodate. I fear that with the growth of our field, scheduling is likely to become more and more difficult. Starting the technical program a day earlier raised problems for some other people, most notably Valerie Miller, the Courses chair: she had to work with courses that overlapped with papers, which caused major scheduling problems.