Real-Time Physics Simulation Forum

In methods,
VoronoiBreakingDemo::voronoiConvexHullShatter
VoronoiBreakingDemo::voronoiBBShatter
j shoud start from i+1 not 1.



-----------

for (i=0; i < numpoints ;i++) {
curVoronoiPoint = points;
btVector3 icp = quatRotate(bbiq, curVoronoiPoint - bbt);
rbb = icp - bbmax;
nrbb = bbmin - icp;
planes.resize(6);
planes[0] = bbvx; planes[0][3] = rbb.x();
planes[1] = bbvy; planes[1][3] = rbb.y();
planes[2] = bbvz; planes[2][3] = rbb.z();
planes[3] = -bbvx; planes[3][3] = nrbb.x();
planes[4] = -bbvy; planes[4][3] = nrbb.y();
planes[5] = -bbvz; planes[5][3] = nrbb.z();
maxDistance = SIMD_INFINITY;
sortedVoronoiPoints.heapSort(pointCmp());
for (j=i+1; j < numpoints; j++) { <<<====================== Here

Your 'fix' breaks the demo, have you tried it?

Oh!

There should be only one fix in the method,
VoronoiBreakingDemo::voronoiConvexHullShatter
j shoud start from i+1 not 1.

Attached is the file with the comment FIX - UPDATE.

[The code has been updated with the most recent one.]

Your "new fix" does not break the demo anymore, because in the demo VoronoiBreakingDemo::voronoiConvexHullShatter is never called.

However if I apply your proposed fix to a demo where voronoiConvexHullShatter is used, it still breaks the demo in a similiar way...

Sorry I was testing this in a different setting
where the convex was only axis-aligned when voronoiConvexHullShatter is called.

Now I have changed my code so it would work on the general setting exactly the same setting like the demo.
It should roughly run 2 times faster than the original demo, by storing the history.

By the way, I know the original code was just for the demo (kindly provided by Alain Ducharme),
so I know it wasn't meant to be as efficient as it could be in the first place.

Thanks.

Also nor my code is as efficient as it could be...

By the way, I have added getVerticesInsidePlanes2.
Attached is the updated header.

The code is updated to re-use the previously computed/found planes.

Note that now the iteration starts at i+1 at // UPDATE - FIX

Indeed!

Thanks.

By the way, the change is reflected only in voronoiConvexHullShatter not in voronoiBBShatter.

The plane computation was incorrect. Attached is the modified code.
In my rough benchmark, it was around 40% faster than the original.