newsKDTree Implementation - News: Windows port (devcpp 5 / Mingw 3.3.1)

Show feedback again
Latest News
On Hold... posted by asmodehn, Sun Jan 3 10:50:52 2010 - 0 replies
Windows port (devcpp 5 / Mingw 3.3.1) posted by asmodehn, Fri Dec 24 13:49:00 2004 - 0 replies
Autotools & CVS cleaning complete posted by asmodehn, Tue Nov 9 11:04:27 2004 - 0 replies
First tests posted by asmodehn, Sat Oct 30 10:54:43 2004 - 0 replies
[Submit News]
[4 news in archive]

Windows port (devcpp 5 / Mingw 3.3.1)

Item posted by Alexandre VINCENT <asmodehn> on Fri Dec 24 13:49:00 2004.

The windows port seems to work well now. I worked around some devcpp 5 beta bugs (in project management).

I have been reading this paper about photonmapping and kdtree search optimisation :

I wonder if this can help me and in which way ?

What I pointed out is :

- In this paper, all the photons are stored before inserting them in the kdtree, so they can do optimisation. This is different from this implementation, because we need to insert photon at any time in the tree (for future merging possibilities), and so have a optimisation algorithm that can run on an already organised kdtree.
Our kdtree must also be usable even if not optimised.
That's why I choose a modified balancement algorithm, but this may change in the future.

- Since a "total balance" algorithm is not used, they need to store the tree with chained pointer. This is well detailed.I have done the same choice after reading this one :
to optimise a bit the balancing algorithm, with som memory considerations

I'd like to have some advice about this. Just mail to kdtree-tech if you have an idea. :)

I also have a bit cleaned the code, on the way to release, planned for today ;)


No messages in Windows port (devcpp 5 / Mingw 3.3.1)


Start a New Thread:

You could post if you were logged in
Show feedback again

Back to the top

Powered by Savane 3.1-cleanup