possible ray vs heightfield bug

foxbat

06-01-2007 08:07:34

Hi.
I'm using the new heightfield optimizations for ODE, and I've discovered a strange potential bug. I'm not sure if this is a problem with the new optimized heightfield code or not, but I don't remember it ever occurring with the unpatched code.

The problem is that if a ray has either it's x or z (or both) components of it's direction set to zero, no collision will be detected. Maybe this has something to do with the AABB of the ray having a 0 length dimension in these cases.

Can anyone confirm this issue?

tuan kuranes

12-01-2007 16:02:25

It does happen in TSM and PLSM2 ?
Did you try to reproduce it on ODE heightfield demo ?

foxbat

28-01-2007 04:11:55

My mistake. It was actually a problem with the ode heightfield code, where collisions would fail on terrain quad grid edges. It looks like you've already fixed this bug though.

xyzzy

25-02-2007 17:27:46

I may have hit this bug. I have my character running around the terrain, and occasionally the character will pass through the terrain. Sometimes this happens while the character is sitting relatively still, so there is not a lot of velocity or acceleration involved.

I believe I am using OgreOde 1.2.0 , Is the bug you described fixed in a later version of OgreOde?