[vtkusers] Recent volume rendering improvements

[Karthik Krishnan]

Hi all:

The empty space leaping on the software ray cast mapper
(vtkFixedPointVolumeRayCastMapper) has been multi-threaded and optimized. On
my laptop the time to compute the datastructures for space leaping of empty
regions in the scalar opacity and gradient opacity transfer functions has
reduced from 5.94s to 2.43s. (core 2Duo), with the use of 2 threads. Please
let me know if there are any issues.


The git diffs are here :

http://vtk.org/gitweb?p=VTK.git;a=commitdiff;h=b02a8044bcea0987a1d1d46778720b94633f41d3

http://vtk.org/gitweb?p=VTK.git;a=commitdiff;h=0040cd160fff70ea24c067ce1ea9e16aa02c6954

http://vtk.org/gitweb?p=VTK.git;a=commitdiff;h=67f285f9fb6fec9ef083528fe0ff2a5e51c98b0a


1. Its multi-threaded now

2. The computation of the scalar opacity and gradient opacity empty regions
are visited in a single pass through the data as opposed to two separate
passes, earlier. There are 3 optimized versions of the passes now and the
appropriate pass is chosen based on whether one or both need to performed.

3. The block size for space leaping can be changed by changing the #ifdef in
the C++ file (VTK_SL_BLK). By default 4x4x4 cells are grouped together to
compute flags indicating whether the region can be skipped or not. You can
if you like play with the number.

Thanks
--
karthik
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20110215/d13e6e83/attachment.htm>