Physically-based modeling: past, present, and future

My name is Demetri Terzopoulos and my co-chair, John Platt, andI would like to welcome you to the panel on Physically-BasedModeling -- Past, Present and Future. I'll start by introducing thepanelists; the affiliations you see listed on the screen aresomewhat out of date. I'm Program Leader of modeling and simulation at theSchlumberger Laboratory for Computer Science in Austin, Texas, andI was formerly at Schlumberger Palo Alto Research. I'll speak onthe subject of deformable models. John Platt, formerly of Cal Tech, is now Principal Scientist atSynaptics in San Jose, California. He will be concentrating onconstraints and control. Alan Barr is Assistant Professor of computer science at CalTech. Last year he received the computer graphics achievementaward. He'll speak about teleological modeling. David Zeltzer is Associate Professor of computer graphics at theMIT Media Laboratory. He will be speaking on interactive microworlds. Andrew Witkin, formerly of Schlumberger Palo Alto Research, isnow Associate Professor of computer science at Carnegie MellonUniversity. He will speak about interactive dynamics. Last but not least, we have with us James Blinn, who of courseneeds no introduction. Formerly of JPL, he is now AssociateDirector of the Mathematics Project at Cal Tech. He says he'll haveseveral random comments to make against physically-basedmodeling. I was also asked by the SIGGRAPH organizers to remind theaudience that audio and video tape recording of this panel is notpermitted. Many of you are already familiar with physically-based modeling,so I will attempt only a very simple introduction to this, in myopinion, very exciting paradigm. Physically-based techniquesfacilitate the creation of models capable of automaticallysynthesizing complex shapes and realistic motions that were, untilrecently, attainable only by skilled animators, if at all.Physically-based modeling adds new levels of representation tographics objects. In addition to geometry -- forces, torques,velocities, accelerations, kinetic and potential energies, heat,and other physical quantities are used to control the creation andevolution of models. Simulated physical laws govern model behavior,and animators can guide their models using physically-based controlsystems. Physically-based models are responsive to one another andto the simulated physical worlds that they inhabit. We will review some past accomplishments in physically-basedmodeling, look at what we are doing at present, and speculate aboutwhat may happen in the near future. The best way to get a feel forphysically-based modeling is through animation, so we will beshowing you lots of animation as we go along. I would like to talk about deformable models, which arephysically-based models of nonrigid objects. I have worked ondeformable models for graphics applications primarily with KurtFleischer and also with John Platt and Andy Witkin. Deformablemodels are based on the continuum mechanics of flexible materials.Using deformable models, we can model the shapes of flexibleobjects like cloth, plasticine, and skin, as well as their motionsthrough space under the action of forces and subject toconstraints. Please roll my Betacam tape. Here is an early example ofdeformable surfaces which are being dragged by invisible forcesthrough an invisible viscous fluid. Next we see a carpet falling ingravity. It collides with two impenetrable geometric obstacles, asphere and a cylinder, and must deform around them. The next clipshows another clastic model. It behaves like a cloth curtain thatis suspended at the upper corners, then released. Here is a simulated physical world -- a very simple worldconsisting of a room with walls and a floor. A spherical obstaclerests in the middle of the floor. You're seeing the collision of anelastically deformable solid with the sphere. Of course, we're alsosimulating gravity. We've developed inelastic models, such as the one you see herewhich behaves like plasticine. When the model collides with thesphere, there's a permanent deformation. By changing a physicalparameter, we obtain a fragile deformable model such as the onehere. This deformable solid breaks into pieces when it hits theobstacle. Deformable models can be computed efficiently in parallel. Thismassively parallel simulation of a solid shattering over a spherewas computed on a connection machine at Thinking Machines, with thehelp of Carl Feynman. Here is a cloth-like mesh capable of tearing. We're applyingshear forces to tear the mesh. The sound you're hearing has beengenerated by an audio synthesizer which was programmed by TonyCrossley so that it may be driven by the physical simulation of thedeformable model. Whenever a fiber breaks, the synthesizer makes apop. Keep watching the cloth; we get pretty vicious with it. Deformable models are obviously useful in computer graphics, butthey are also useful for doing inverse graphics; that is to say,computer vision. For example, here we see an image of a garden variety squash.Using a deformable tube model, we can reconstruct a threedimensional model of the squash from its image, as shown. Once wehave reconstructed the model from the image, we can rotate themodel to view it from all sides. You can see, we have captured afully three dimensional model from that single, monocular image.That's a basic goal of computer vision. Kurt Fleischer, Andy Witkin, Michael Kass, and I used thisdeformable model based vision technique to create an animationcalled <i>Cooking with Kurt.</i> We wanted to mix livevideo and physically-based animation in this production. You seeKurt entering a kitchen carrying three vegetables. We captureddeformable squash models from a single video frame of the realsquashes sitting on the table -- this particular scene right here.Now the reconstructed models are being animated usingphysically-based techniques. The models behave like very primitiveactors; they have simple control mechanisms in them that make themhop, maintain their balance, and follow choreographed paths. Thecollisions and other interactions that you see are computedautomatically through the physical laws, and they look quiterealistic. It's difficult to do this sort of thing by hand, even ifyou're a skilled animator. This second tape will show you some of the physically-basedmodeling we're up to now at the Schlumberger Laboratory forComputer Science. Keith Waters and I are working on interactivedeformable models. We're now able to compute and render deformablemodels in real time on our Silicon Graphics Iris 240 GTX computer.For example, here is a simulation of a nonlinear membraneconstrained at the four corners and released in a gravitationalfield. Watch it bounce and wiggle around. Here you're seeing a physically-based model of flesh. It's athree dimensional lattice of masses and springs with musclesrunning through it. Again, this is computed and displayed in realtime. You can see the muscles underneath displayed as red lines.They're fixed in space at one end and attached to certain nodes ofthe lattice model at the other end. By contracting the muscles wecan produce deformations in this slab of -- whale blubber, if youwill. We did this simulation as an initial step towards animatingfaces using deformable models as models of facial tissue. And ofcourse, the muscle models make good facial muscles. The next clip will demonstrate real time, physically-basedfacial animation on our SGI computer. Here we see the latticestructure of the face. Let's not display all of the internal nodesso that we can see the epidermis of the lattice more clearly.There. Now we're contracting the zygomatic muscle attached to oneedge of the mouth -- now both zygomatics are contracting to createa smile. The muscles inside the face model are producing forceswhich deform the flesh to create facial expressions. Now the epidermis polygons are displayed with flat shading. Nextwe contract the brow muscles. Here the epidermis is being shadedsmoothly. Finally, we relax the muscles and the face returns tonormal. An important reason for applying the physically-based modelingapproach to facial animation is realism. For instance, the facialtissue model automatically produces physically realistic phenomenasuch as the laugh lines around the mouth and the cheek bulges thatyou see here. Keith videotaped this animation off of our machine only lastweek. Our next step will be to develop control processes tocoordinate the muscles so that the face model can create a widerange of expressions in response to simple commands. Keith's priorwork on facial animation, published in SIGGRAPH 87, showed how onecan go about doing this using muscle model processes. Beyond musclecontrol processes, we're also interested in incorporating vocodermodels -- that is, physically-based speech coding and generationmodels, so that this face can talk to you. The tape will end soon, so I'll release the podium to Dr. JohnPlatt, who will talk about constraint methods and control. Thankyou.