If I recall correctly the collision mesh cant have submeshes, or they can but only first is applied. This means that if you put the same mesh as collision mesh only the first submesh will do collision. So the "area" of the mesh that the first material affects too, is the first submesh. For complex meshes that has many submeshes to set own textures you can simply make a separate collision mesh of your model. When the mesh is done, combine all the submeshes in your editor and export that separately, upload and set as that collision mesh.
Example: you have platform where you want to walk, with "risen side part" submesh for best visual looks with its own texture. If you set the same mesh you'll fall trough the platform part but can walk on the "sides" if its the first submesh. Hope this helps. I dont think its too hard to make one mesh out of a complex one as a separate collision mes in any editor :) Best regards, Jonne Nauha realXtend developer http://www.realxtend.org/ http://www.evocativi.com/ On Mon, May 31, 2010 at 9:21 AM, Toni Alatalo <[email protected]> wrote: > Gustavo Alberto Navarro Bilbao kirjoitti: > >> Yes, I know that solution: to create a big and invisible meshe over the >> other ones, but, in my opinion that is the kind of provisional solution than >> in spanish we call "una chapuza", sorry but I don't know the english >> definition. >> > > No not as a separate object, but one(s) that you define as the *collision > mesh* of the visual object. In the rex props tab, IIRC next to the where you > assign the visual mesh, you can define that collision mesh used too. > > It is not a hack or some sort of cheap solution, it is how realtime content > with collisions has usually been authored. > > > That solution is good for visual aspects and no complex meshes, but if you >> really would like to create and "inmersive" world and not only a "virtual >> walktrought" one, that can't be the final solution, specially if we are >> thinking about scripts to test wheels, engines and other perfomances in >> simulations. >> > > Err, there is no limit for the complexity or anything set by this route of > having the collision object definition separate. Quite the opposite, it > allows for more complexity, 'cause the visual mesh can be made for best > visuality and the collision mesh for best collision features. > > Having the fallback where the visual geom is used also for collisions, when > a separate collision mesh is not provided, is useful for quick/simple things > but exactly more complexity is where you may need this feature that it is > possible to use a separate one too. > > Alberto >> > > ~Toni > > 2010/5/31 Toni Alatalo <[email protected] <mailto:[email protected] >> >> >> >> >> Gustavo Alberto Navarro Bilbao kirjoitti: >> >> I think that it would be very important to fix the issue of >> physics in the meshes, when they have more materials, to be >> cease to be "phantoms", >> >> >> I suppose you know the current solution, from the earlier talks? >> Use a collision mesh without several materials, for the visual >> objects that have many materials. That's the reason why it has >> been implemented like it is: collision meshes are often authored >> separately, and as they are invisible anyway, there is no reason >> for them to have materials, so the physics mesh creator didn't >> need to support that. If you don't care / need to make a different >> geom for the collisions (often many visual details are irrelevant >> or even harmful for proper collisions), you can just make a copy >> of the mesh where remove the mats in your modelling app. >> >> But I agree that for the fallback of using the visual mesh for >> collisions too when a separate collision mesh is not provided it >> would be a good idea for it to handle all the submeshes (material >> indexes become submeshes in ogre). Probably quite simple to add in >> rexode where the geom for ode is created. >> >> ~Toni >> >> >> -- http://groups.google.com/group/realxtend >> http://www.realxtend.org <http://www.realxtend.org/> >> >> >> >> -- >> http://groups.google.com/group/realxtend >> http://www.realxtend.org >> > > -- > http://groups.google.com/group/realxtend > http://www.realxtend.org > -- http://groups.google.com/group/realxtend http://www.realxtend.org
