I would agree with Xiaohui Bie, this si something that needs great improvement with CS, so much so that my team got frustrated with CS and we will probably migrate to a new engine. Definitely needs a full implementation of Bullet or a newer physics library to handle dynamic collision detection, that is what most games need nowadays and makes development a lot easier if this was integrated right.
Brad Nelson Play-Em > Biography > > > My name is Xiaohui Bie. I am a second year graduate student of Institute > of Software, Chinese Academy of Sciences (ISCAS). My major is Computer > Graphics and my research focuses on physics-based simulation. I have been > researching on collision detection for nearly two years and obtain a > strong understanding in this area. I am familiar with many good physics > libraries such as opcode, bullet and solid. My experience in C/C++ is > almost 5 years and I¡¯m proficiency in OpenGL coding. > I have led some projects, such as Manufacturing production management > software MRP, Project Management software MRP. I hope this chance can lead > me to Crystal Space world and I¡¯m eager to contribute. > > > > Project/ideas > Crystal Space uses two different systems to cope with collision and > dynamics respectively, which may bring developers especially the recruits > some unconvenience , I was once a recruit, so I especially know how > necessary an unified collision and dynamics system is. I think some modern > physics libraries are worthy of study, such as bullet, which allows us to > do only collisions detection without the dynamics. To unify the two into > one set, we can build two interface like the bullet, one is collisionworld > and the other is dynamicsworld. Dynamicsworld inherits from > collisionworld. The collisionWorld is the interface for the collision > detection., while the dynamicsworld is the interface for several dynamics > implementation. If we only need collision detection, just like the bullet > collisionintefacedemo does, use the collisionworld is enough. Because > collisionworld is above dynamicsworld, if you want to build a dynamic > world, collisonworld is also needed to be include. > Further more, because I have been studied in this area for a long time, I > like to share my experiences with others, so I would like to introduce > something new to the collision detection of Crystal Space, such as the > collision detection of deformed objects, although it may be involved in > some libraries, for the self-collision, they are also blank there. But > self-collision occurs very frequently in our animation scenes, such as > cloth simulation, which inevitably involves self-collisions. In fact, > there are great progress in this area recently, such as¡¾Volino et al > 06¡¿,¡¾Sud et al 06¡¿,¡¾Govindaraju et al 05¡¿. Any one of them can attain > a good result. Since the BVHs refit algorithm for deformed objects is very > mature and fast, so all we need to do is the introduction of Chromatic > Decomposition method in¡¾Govindaraju et al 05¡¿ or Discrete Voronoi > Diagrams method in ¡¾Sud et al 06¡¿. > > > References > ¡¾1¡¿N. Govindaraju, D. Knott, N. Jain, I. Kabul, R. Tamstorf, R. Gayle,M. > Lin, and D. Manocha, ¡°Interactive collision detection between deformable > models using chromatic decomposition,¡± ACM Trans.Graph. (SIGGRAPH Proc.), > vol. 24, no. 3, pp. 991¨C999, 2005. > > ¡¾2¡¿A. Sud, N. Govindaraju, R. Gayle, I. Kabul, and D. Manocha, ¡°Fast > proximity computation among deformable models using discrete Voronoi > diagrams,¡± ACM Trans. Graph. (SIGGRAPH Proc.), vol. 25,no. 3, pp. > 1144¨C1153, 2006. > ¡¾3¡¿P. Volino and N. Magnenat-Thalmann, ¡°Resolving surface collisions > through intersection contour minimization,¡± ACM Trans.Graph. (SIGGRAPH > Proc.), vol. 25, no. 3, pp. 1154¨C1159, 2006. > ¡¾4¡¿physics simulation forum http://www.bulletphysics.com > > 2009-03-27 > > > > YOURS SINCERELY, > > Xiaohui Bie > CG Group, State Key Laboratory of Computer Science, > Institute of Software,Chinese Academy of Sciences(ISCAS) > Tel: (+86)15901521335 > Email: [email protected] > ------------------------------------------------------------------------------ > _______________________________________________ > Crystal-main mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/crystal-main > Unsubscribe: > mailto:[email protected]?subject=unsubscribe > ------------------------------------------------------------------------------ _______________________________________________ Crystal-main mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/crystal-main Unsubscribe: mailto:[email protected]?subject=unsubscribe
