Just a quick post to say that currently my attempt at implementing Alpha Hulls is a long way from running. I have also located a backup of my old mesh/adjacency although I have yet to gather any statistics on its run time.
So I guess I need to just keep plugging along with my implementation, currently blocked out as
1. Look for hard edges (connected face normals seperated by > 70 degrees)
2. Compute the alpha prisms
3. Create a lookup for vertices sorted by position into a grid with cells (2*m_alpha) x (2*m_alpha) x (2*m_alpha) This is really just an optimisation to reduce searches to a subset
4. Sort edges into sets of directed strips
5. Check for contacts
6. Add and remove edges identified by contacts.
Im still not sure that I fully understand how to check for contacts such that it resolves to the correct point so maybe you have some thoughts on it Tyrone, either way it could be something to discuss.
No comments:
Post a Comment