Hi: As Jorrit said, cloth simulation is definitely a valid idea for us. You seem to have the knowledge necessary to complete this project, so I would recommend writing up a formal proposal and submitting it via the Google Summer of Code submission system.
Some thoughts: Thus Spoke Мая Милушева: > Hello everybody! I’m Maya Milusheva and I want to apply for a GSoC > project. > I'm interested in developing cloth simulation and a few weeks ago > Marten Svanfeldt (thebolt) suggested me to use the Hierarchical > Position Based Dynamics approach based on > http://www.matthiasmueller.info/publications/hpbd.pdf article. > > I haven't read this article, but it seems interesting. Another article I would recommend, as it is really interesting, is from SIGGRAPH this past summer: http://www.cs.cornell.edu/~srm/publications/SG08-knit.html It's about simulating knitted materials, which many types of clothing appear under. It might be a little too specific for what you want to do, but nonetheless, it could be an interesting approach to simulating clothing. A disadvantage, however, would be that it would probably have to be pre-baked, but you might be able to take some of the ideas and come up with a real time version. > I've been studying about HPBD for the last few weeks and I came up > with a rough design. I'll explain the project and the algorithm in > more details in my proposal. However here's briefly what I want to > achieve with this project. > > What is good about HPBD is that it can handle pretty much all > sorts of constraints you put in it (for bending, for distance, for > keeping a constant volume, etc.). That is why I want to create a > flexible system that will be able to simulate different dynamics on > objects just by adding various constraints. I'll have a manager class > csHPBDSimulationManager (just an example name) that will hold all the > active simulations and take care of updating them, adding/removing > simulations and so on. csHPBDSimulation class will hold a list of all > constraints, add/remove constraints, hold the mesh for the simulation > etc. It will also hold a pointer to csHPBDDynamics which is the most > important class. csHPBDDynamics is where most of the work is done that > is creating the hierarchy for the multigrid, project the constraints, > update the velocities and positions and others. And also I'll create > an abstract class csHPBDConstraints so that each time when the > programmers want a new sort of constraint they will just extend this > class and create a new constraint function. > Ok, this sounds great. One question I have, though is whether this will all run in real time? > So basically when the project is done the programmers should only > need to create a few sorts of constraints by extending the abstract > class (for example for bending and distance) then create a new object > of csHPBDSimulation and pass to it the constraints and mesh and then > add the new csHPBDSimulation to the csHPBDSimulationManager who will > start updating the simulation. I'll make a basic set of constraint > functions that can be used directly by the programmers to ease them. > As a demonstration I'll create cloth simulation that can be used as an > example for other simulations. > > Let my know if it's a good idea and if it's suitable for a GSoC > project. Should I change something, should I submit my proposal? If > you have any questions about me or the project I'll be happy to > answer. > Any feedback is welcome. > > Definitely a good idea. :) ~Scott ------------------------------------------------------------------------------ _______________________________________________ Crystal-main mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/crystal-main Unsubscribe: mailto:[email protected]?subject=unsubscribe
