Im hardly an expert, im just getting started actually… But wouldn’t it be a lot easier if per polygon collision detection could be done on the GPU? I know its not feasible w/ today’s hardware architectures… but if you could just check for 2 polygon’s collisions inside of OGL, and it return the point of collision??.. It’d kinda merge the graphics and physics side of programming… but it could save framerates as opposed to calculating collisions in software??
This gets suggested about once a month in this forum. Read previous posts in this forum for some discussions on this topic.
jeeze sorry didn’t mean to be redundant
If you want ray-to-polygon intersection, the select or feedback buffer can help you.
But if you want polygon-to-polygon intersection, it is generally not possible or hard to implement.
Why waste the GPU’s time in doing this? The GPU should be used to enhance your scene instead.
When designing your game/application, you should always balance the tasks of your GPU(s) and CPU (s) so if you think you want to do this, check out the cg pages. Someone has made an attempt, but it’s only for show.
Why waste the GPU’s time in doing this?
It’s not really a waste of time if the application is not realtime.If CPU is already overloaded, it can be a good thing to use GPU in order to increase overall processing power.
If you have a well designed scene management system this problem is not that hard as you might expect. CPU overload you mentioned is oftenly caused by such problems. So you want to overload the GPU…