>From what I understand you can in broomstick. I implemented a version of ccd in my game that splits the screen into 9, and only objects that shared sections or bordered on sections were tested for collisions. This seemed to sort out the high number processor cycles. Just means a bit more coding. I held all of the code in a singleton collision class and called it's functions from the game manager, which passed in all of the arrays that needed testing.
Good luck on your project Liam On Mar 18, 5:06 pm, Arkadianen <[email protected]> wrote: > Thanks for the book hint. > Josh did you solve the productivity issue? Collision detection can > take a lot of processor resources. Can we calculate it somehow in GPU? > > On Mar 17, 7:36 pm, Josh Beckwith <[email protected]> wrote: > > > > > I just posted the source for a project I'm working on, and I think you > > might find some of it useful. > > > I've set up a really simple ball-firing class that spawns a ball and > > gives it direction and speed based on where the camera is looking. > > It's for Away3D4, but if you're not using that I'm sure you could > > replace a couple of methods and it will work. > > >http://groups.google.com/group/away3d-dev/browse_thread/thread/519b41... > > > On Mar 17, 9:04 am, pokey <[email protected]> wrote:> Many thanks. > > I'd like to eventually have non-constant speed > > > projectiles, but I think nailing down the constant speed case will be > > > a good first step. > > > > Coincidentally, I've been glancing at that book's preview via google > > > books, and have it ordered on Amazon. > > > > On Mar 17, 6:45 am, Michael Iv <[email protected]> wrote: > > > > > There is an issue with detecting collision on the bodies travailing very > > > > fast.In a regular way it may occur that at the next collision test the > > > > object (bullet ) is already past the collision point. For this CCD is > > > > the > > > > solution .Ray is a simplest solution however. You cast it in the > > > > direction > > > > of the bullet path.You get the hit ( the distance to the collision > > > > point) > > > > .Assuming that the travel time is CONSTANT ray model works fine > > > > .Because in > > > > this case you get the exact time of the hit and have an ability to > > > > predict > > > > the impact.If the speed is changing or the impacted object is dynamic > > > > you > > > > should better got CCD solution.It really depends on if you want to have > > > > a > > > > physical behavior of the projectile becasue in this vase you should take > > > > into account (acceleration,gravity ,drag forces etc) or just a simple > > > > bullet > > > > constant speed model which would be pretty easy to simulate."Physics for > > > > Game Programers" book is very nice resource to learn this stuff. > > > > > On Thu, Mar 17, 2011 at 3:34 PM, Arkadianen <[email protected]> > > > > wrote: > > > > > How CCD differs from collision iteration? > > > > > Can we calculate collisions in GPU ? What other operations can be made > > > > > in GPU? > > > > > > On Mar 16, 10:05 pm, Michael Iv <[email protected]> wrote: > > > > > > Usually in fps shooters ray is used to get hit position . Another > > > > > > method > > > > > for bullets hit test is CCD ( continuous collision detection ) . It > > > > > can be > > > > > used with rays or on it's own . Google for the exact implementation . > > > > > If I > > > > > recall right Seb Lee had an example on his blog . Also check JigLib > > > > > may be > > > > > it has got built in CCD > > > > > > > Sent from my iPhone > > > > > > > On Mar 16, 2011, at 9:02 PM, pokey <[email protected]> wrote: > > > > > > > > Hi there, > > > > > > > > I'm new to Away3D, and trying to create an oh so simple little > > > > > > > shooting gallery: it will have a plane with some cubes scattered > > > > > > > on > > > > > > > it, and you will use the mouse as a crosshair to fire projectiles > > > > > > > at > > > > > > > the cubes, from a vantage point hovering over the plane. > > > > > > > > My hangup is getting the projectile firing, and testing whether it > > > > > > > hits a target (the cubes, or the plane itself). I'll have some > > > > > > > function defining the position of the projectiles as a function of > > > > > > > time and where the crosshairs were aimed when they were released. > > > > > > > At > > > > > > > this point, I could just have iterate through every cube (and the > > > > > > > plane) at each frame testing if one of the projectiles has > > > > > > > collided > > > > > > > with it. I'm wondering if anyone could suggest a simpler way to > > > > > > > proceed utilizing some Away3D libraries? The getIntersect() > > > > > > > method in > > > > > > > Ray.as looks promising, though it's not completely clear to me > > > > > > > what > > > > > > > this does (nor can I find much documentation). > > > > > > > > Cheers, > > > > > > > P > > > > > -- > > > > Michael Ivanov ,Programmer > > > > Neurotech Solutions Ltd. > > > > Flex|Air |3D|Unity|www.neurotechresearch.comhttp://blog.alladvanced.net > > > > Tel:054-4962254 > > > > [email protected] > > > > [email protected]
