HOME | DD

Description
It took damn near a week to get the tail to behave properly.

ODE for physics and OGRE for general graphics (even though I had to write all the lightmapping and shadowing myself *grumblegrumble*).

But anyway... here's a little preview of what my game looks like right now. I'm finally to the point where I'm pleased with Lily's tail in the game. I also decided to go ahead and take the advice of some fellow deviants on making her tail thicker and longer.

And now for some random rantings...

Getting the world position and orientation of a specific Ogre::Bone is a lesson in frustration. But that's the coordinate space that EVERYTHING IN ODE uses. Setting said position is just as bad. In the end I decided to take the code I'd written to GET the world position and just translate it in the world coordinate space (because for some reason that works just fine, even though it's impossible to get the world coordinates or a matrix to convert to/from world coordinates) to move it to the new world coordinate. AAARGH.

And on the topic of ODE... ODE doesn't do continuous collision detection. Basically, it can test for collisions before and after you move something, but it won't extrude out a shape to test collisions across its movement path. So if your object is moving fast enough to end up completely on the other side of another object in one iteration, then it'll never see the collision.

I'm doing everything multithreaded here, which is a hell of a lot more bookkeeping. It's a tremendous pain in the ass, but the performance on multi-core systems is amazing. I also have a feeling that OGRE sets up the render queue (OpenGL) and waits for it to finish all in one command, so there's plenty of idle CPU time while the GPU works. The point that I'm making is that, even on a single-core system, there are possible advantages of the multithreaded organization.

And stuff.

Description
It took damn near a week to get the tail to behave properly.

ODE for physics and OGRE for general graphics (even though I had to write all the lightmapping and shadowing myself *grumblegrumble*).

But anyway... here's a little preview of what my game looks like right now. I'm finally to the point where I'm pleased with Lily's tail in the game. I also decided to go ahead and take the advice of some fellow deviants on making her tail thicker and longer.

And now for some random rantings...

Getting the world position and orientation of a specific Ogre::Bone is a lesson in frustration. But that's the coordinate space that EVERYTHING IN ODE uses. Setting said position is just as bad. In the end I decided to take the code I'd written to GET the world position and just translate it in the world coordinate space (because for some reason that works just fine, even though it's impossible to get the world coordinates or a matrix to convert to/from world coordinates) to move it to the new world coordinate. AAARGH.

And on the topic of ODE... ODE doesn't do continuous collision detection. Basically, it can test for collisions before and after you move something, but it won't extrude out a shape to test collisions across its movement path. So if your object is moving fast enough to end up completely on the other side of another object in one iteration, then it'll never see the collision.

I'm doing everything multithreaded here, which is a hell of a lot more bookkeeping. It's a tremendous pain in the ass, but the performance on multi-core systems is amazing. I also have a feeling that OGRE sets up the render queue (OpenGL) and waits for it to finish all in one command, so there's plenty of idle CPU time while the GPU works. The point that I'm making is that, even on a single-core system, there are possible advantages of the multithreaded organization.

And stuff.