[yt-svn] New issue 344 in yt: Volume rendering refactoring

[Bitbucket]

--- you can reply above this line ---

New issue 344: Volume rendering refactoring
https://bitbucket.org/yt_analysis/yt/issue/344/volume-rendering-refactoring

Matthew Turk / MatthewTurk on Sat, 26 Nov 2011 14:50:11 +0100:

Description:
  Right now (2.3), the volume rendering is set up to be very tightly coupled.  The partitioned grid has a ray integration method, a sample values method, and then these call the transfer evaluation method on the transfer function.

While in principle all of these items are necessary and important, they lead to an overgenerality in some respects and overspecificity in others.  For instance, there is no optimized method of accumulating values rather than evaluating transfer (overgeneral) and there is no method of *not* subsampling rays inside a cell (overspecificity).

The volume rendering subsystem should be refactored to set up:

 * Ray walking routines that accept arbitrary segments of data in space, not necessarily PartitionedGrid objects, and to walk through them (with masking)
 * Arbitrary ray sampling routines.  For instance, this could be passing in a function pointer to a sampling (or non-subsampling) method that picks up values and integrates a plane *or* even talks back to the underlying data.  This should support for instance laying down values in cells as well as picking up values from cells.
 * More flexible image accumulation and transfer functions, for things like hard surfaces and lighting.

Responsible:
  MatthewTurk
--

This is an issue notification from bitbucket.org. You are receiving
this either because you are the owner of the issue, or you are
following the issue.