Jump to content

  • Log In with Google      Sign In   
  • Create Account

Banner advertising on our site currently available from just $5!

1. Learn about the promo. 2. Sign up for GDNet+. 3. Set up your advert!


Member Since 21 Mar 2012
Offline Last Active Yesterday, 09:42 AM

Posts I've Made

In Topic: Calc maximum INNER AABB

19 March 2015 - 06:53 AM

No need to get snippy

What wrong I said? Just question for ideas...

Means, that you will most likely not find a prepared, ready to use solution.

I understand it.

overlapping AABBs allowed ? Min/max size of AABB ? Are AABB allowed to include vertices ? Min/max number of AABBs per Mesh ?

overlapping - yes; size - as max as possible; including vertices is allowed, but not too much(5-10% by all); number of aabbs - from 1 to 4-5

navigation mesh

yep, it's really good idea.
get 3 planes, at x, y, z axis aligned and it will be contour of x, y, z projection of mesh.

hard problem here is:
How to simplify projected countour(3d mesh projection to 2d plane by orthogonal camera, using software rasterizer) to minimum vertices?

In Topic: Calc maximum INNER AABB

18 March 2015 - 12:10 PM

it sounds not like a trivial problem.


any other ideas? )

may be anyone has sample of https://mediatech.aalto.fi/~ari/Projects/OSPS/ ?

In Topic: Calc maximum INNER AABB

16 March 2015 - 11:33 PM

The idea:

I don't really understand your idea...
You suggest to reduce AABB, checking every corner of AABB?

Not aabb's but something better with similar triangle counts.

It's good idea, but it looks difficult to implement.

In Topic: Dynamic Octree, Loose Octree is the best solution ?

15 March 2015 - 11:34 PM

Why loose octree is better for a dynamic octeee ?


because when objects move at small distance, octree not reconstrunct, right?

In Topic: Calc maximum INNER AABB

15 March 2015 - 11:32 PM

For what use case you need those?

for occlusion culling.


If the model can be any arbitrary shape, then an infinite number of infinitely small boxes would be "best." You need to better describe constraints on "best fill" to get a better answer.

If I say "maximum aabb", it does not mean that I need small boxes, ok?

if model very complex, it needs in several boxes, than one.