Manta Interactive Ray Tracer Development Mailing List

Text archives Help


Re: [Manta] Re: Manta: Direct isosurfacing using octrees


Chronological Thread 
  • From: Aaron Knoll <knolla@cs.utah.edu>
  • To: Shreekanth Pavar <shreekanth.pavar@ucl.ac.uk>
  • Cc: MANTA <manta@sci.utah.edu>
  • Subject: Re: [Manta] Re: Manta: Direct isosurfacing using octrees
  • Date: Sun, 21 Oct 2007 12:18:16 -0700
Hi Shree,

It's been a while since I used the OctreeVolume code. It certainly isn't perfect; it was research code that was never really designed for robust end-user support. For this I apologize...

On the other hand, for truly large data octree volumes are still probably a practical solution, more so now that multicore CPU's are more powerful. Out of curiosity how big is your data?

First of all, did the octree volume (octvol_build) builder work, without any seg faults or errors?

OctNode& node = octdata->get_node(depth, node_index);

That's the basic point-location lookup routine for a node. Usually when this fails, it's because the data isn't built correctly.

Here's what I would try:

1) place a "cerr<<" with the index of the octree node it thinks it's trying to access
2) run in debug mode and see what's happening
3) send me your octree volume data. This could be difficult because I am working behind a firewall at Intel, but if necessary we'll work something out :)

-Aaron

On Oct 21, 2007, at 11:46 AM, Shreekanth Pavar wrote:

Hi Aaron

I'm currently trying to render some of our datasets using the Octree Direct
Isosurfacing code however I get a segmentation fault in the (recursive)
function:

IsosurfaceOctreeVolume::single_traverse_node

when trying to retrieve the first data node at this line:

OctNode& node = octdata->get_node(depth, node_index);

The arguments are both supplied hardcoded to the value 0 in the calling function
(IsosurfaceOctreeVolume::single_intersect) so my guess is that there is
something wrong with my Octree data itself even though the octree volume
builder appeared to create it ok from the .raw volume data file that I
supplied.

Do you know if there is any way to examine the data directly for faults? Also,
are any sample octree volumes available which I could use to confirm that the
problem lies with the data and not in the code?

Many Thanks


--
Shree Pavar
Centre for Computational Science, UCL
http://www.chem.ucl.ac.uk/ccs/




Archive powered by MHonArc 2.6.16.

Top of page