I am working on some rendering of data using voxel datapoints. There is
something I don't really understand going on with bounding and translate limits
or possibly a bug or overflow.

I am on ubuntu 10.04 with povray version:
Persistence of Vision(tm) Ray Tracer Version 3.6.1 (Debian  (x86_64-linux-gnu-g+
+ 4.3.3 @ x86_64-pc-linux-gnu))

Take this POV file here: http://pastebin.com/Bp5H9swS [test.pov]

It loads two rows of cubic voxels, however in rendering it with:
povray -Q0 test.pov

It looks like: http://i.imgur.com/nBwb2.png

Notice the gabs between the blocks, they should not be there based on the data.

However if rendered with:

povray -Q0  -MB test.pov

The expected result is obtained here: http://i.imgur.com/odoR8.png

Notice the smooth row of blocks.

Fair enough seems like bounding is the issue, obviously rendering is MUCH slower
with -MB . I would really like a way to optimize to a more +MB speed. The really
odd part though is that if I simply change the translate positions to use
smaller numbers as in this .pov file:
http://pastebin.com/jkm9fX3U [test2.pov]

Changing a value like this:
object { MyMesh translate <7458344, 721, 13692132> }

to this:
object { MyMesh translate <8344, 721, 2132> }

and then run:

povray -Q0 test2.pov

I get the correct result with the +MB performance here:

http://i.imgur.com/yKJme.png

Is there something I can do settings wise or other options (even compile time)
to resolve this?

Thanks.

Post a reply to this message