CS-184: Computer GraphicsLecture #21: Physically Based Animation IntroProf. James O’BrienUniversity of California, BerkeleyV2009-F-21-1.02Today•Introduction to Simulation•Basic particle systems•Time integration (simple version)Tuesday, November 24, 20093•Generate motion of objects using numerical simulation methodsPhysically Based Animationxt+Δt= xt+ Δt vt+12Δt2atgv4Physically Based AnimationComputer Graphics Proceedings, Annual Conference Series, 2003Figure 8: An explosion under an immobile arch.Figure 9: An explosion between a group of immobile pillars.Foster, N., and Metaxas, D. 1997. Modeling the motion ofa hot, turbulent gas. In Proceedings of ACM SIGGRAPH 97,181–188.Jensen, H. W., and Buhler, J. 2002. A rapid hierarchicalrendering technique for translucent materials. In Proceedingsof ACM SIGGRAPH 2002, 576–581.Lamorlette, A., and Foster, N. 2002. Structural modelingof natural flames. In Proceedings of ACM SIGGRAPH 2002,729–735.Lee, H., Kim, L., Meyer, M., and Desbrun, M. 2000. Mesheson fire. In EuroGraphics 2000 Workshop on Animation.Lokovic, T., and Veach, E. 2000. Deep shadow maps. InProceedings of ACM SIGGRAPH 2000, 385–392.Martins, C., Buchanan, J., and Amanatides, J. 2002. Ani-mating real-time explosions. The Journal of Visualization andComputer Animation 13, 2, 133–145.Mazarak, O., Martins, C., and Amanatides, J. 1999. Animat-ing exploding objects. In Graphics Interface 99, 2 11–2 18.McGrattan, K. B., et al. 2002. Fire dynamics simulator (ver-sion 3) technical reference guide. Tech. Rep. NISTIR 6783,2002 Ed., National Institute of Standards and Technology.Melek, Z., and Keyser, J. 2002. Interactive simulation of fire.In Pacific Graphics 2002, 431–432. Presented in poster session.Meyer-Arendt, J. R. 1984. Introduction to classical and modernoptics. Prentice-Hall.Neff, M., and Fiume, E. L. 1999. A visual model for blast wavesand fracture. In Graphics Interface 99, 193–202.Nguyen, D. Q., Fedkiw, R. P., and Jensen, H. W. 2002. Phys-ically based modeling and animation of fire. In Proceedings ofACM SIGGRAPH 2002, 721–728.O’Brien, J. F., and Hodgins, J. K. 1999. Graphical model-ing and animation of brittle fracture. In Proceedings of ACMSIGGRAPH 99, 137–146.O’Brien, J. F ., Bargteil, A. W., and Hodgins, J. K. 2001.Graphical modeling and animation of ductile fracture. In Pro-ceedings of ACM SIGGRAPH 2001, 291–294.Reeves, W. T. 1983. Particle systems — a technique for modelinga class of fuzzy objects. In Proceedings of ACM SIGGRAPH83, 359–376.Rushmeier, H., Hamins, A. , and Choi, M. Y. 1995. Volumerendering of pool fire data. IEEE Computer Graphics & Ap-plications 15, 4 (July), 62–67.Stam, J., and Fiume, E. L. 1995. Depicting fire and other gaseousphenomena using diffusion processes. In Proceedings of ACMSIGGRAPH 95, 129–136.Stam, J. 1999. Stable fluids. In Proceedings of ACM SIGGRAPH99, 121–128.Yngve, G. D., O’Brien, J. F., and Hodgins, J. K. 2000. Ani-mating explosions. In Proceedings of ACM SIGGRAPH 2000,29–36.Figure 10: Cutaway view of the explosion shown in Figure 7.715Animating FractureJames F. O’Brien Jessica K. HodginsMay 31, 2000The task of specifying the motion of even a simple animated object, likea bouncing ball, is surprisingly difficult. In par t, the task is difficult becausehumans are very skilled at observing mo vement and quickly detect motion that isunnatural or implausible. Additionally, the motion of many objects is complexand specifying their movement requires generating a great deal of data. Forexample, cloth can bend and twist in a wide variety of ways, and the breakingbunny statue shown in Figure 1.a involves many hundreds of individual shards.Three primary techniques are used to generate synthetic motion: keyfram-(a)(b)Figure 1: These images s how the results of using our technique to simulate thebehavior of (a) a hollow, ceramic bunny as it is stuck by a hea vy, fast-movingweight, and (b) a slab of simulated glass that has been shattered by a heavyweight.1Figure 5: This image sequence shows frames from an animation of a pair of objects colliding with each other. Eachobject is a hybrid simulation that incorporates a rigid and a deformable (modal) component.Figure 6: These images shows how constraints can beused to deform objects. The object on the left of each im-age shows the object prior to deformation, and the rightobject shows the results after the red constraint pointshave been moved.Figure 7: These images are screen shots from an applica-tion running natively on a Sony PlayStation2. The yellowcircle highlights the cursor that the user is using to pokeand pull an elastic figure.around 10 points on the model can be constrained in real-time on a moderate speed computer (300 MHz PentiumII or Sony Playstation2). A limit is reached because thesolutions to equation (13) and equation (15) require a rel-atively expensive computation of singular value decom-positions, which cannot be calculated in real-time oncethe matrices become too large.We have created several animations (see supplementalmaterials) demonstrating this system, each simulated in-teractively for moderately complex objects. The resultsappear plausible, and resemble animations that might besimulated using more straightforward but more compu-tationally expensive methods. The bottlenecks in hybridmodal/rigid-body simulation are collision detection andsolving the linear program for the constraints. To reducethe computation used in solving the linear program, theextent of contact point clustering may be tweaked to sac-rifice accuracy for speed. Figures5 and 8 show objectsinvolved in collisions with a ground plane and each other.Figure 8: A sequence of images showing the StanfordBunny model bouncing off a ground plane.As with other methods based on tetrahedral finite el-ements, we can embed high-resolution or non-manifoldsurfaces inside a tetrahedral volume model. The bene-fits of this technique are that the surface shading and tex-turing can be specified independently from the dynam-ics, and poorly constructed “polygon-soup” models maybe used. Both the brain model in figure 1, an extremelycomplex object, and the “dodo” model in figure5, a non-manifold object, are modeled in this way. The “dodo”model also demonstrates non-uniform material proper-ties: the legs and beak are made of a stiffer material thanthe rest of the body.5 ConclusionsModal analysis has been shown to be a useful tool for in-teractively producing realistic
View Full Document
Unlocking...