Re: BACK_TO_FRONT sorting and 'Inverse Singular Matrix'

New Message Reply Date view Thread view Subject view Author view

Angus Dorbie (dorbie++at++sgi.com)
Wed, 06 Jan 1999 17:46:32 -0800

You sure there isn't a culprit DCS in the graph with a singular matrix,
for example shadows on the ground with a zero scaling factor in Z?

FYI this would cause z fighting which wouldn't be solved by sorting.
It's a different class of problem.

What machine are you using?

Also for sort to look good you need the objects with transparency to be
split up at the granularity of potential occlusion. Sorting only happens
on geosets so objects which potentially occlude each other need to be is
separate geosets. You chould also consider the centroid & size of these
objects, a simple depth sort will not work if objects overlap
significantly on some axis w.r.t. their rize

Cheers,Angus.

Greg Harrison wrote:
>
> When we switched our program from PFSORT_BY_STATE to PFSORT_BACK_TO_FRONT
> (because of overlapping MS_ALPHA transparency), we get a pf error about
> 'Inverse Singular Matrix' when there are many objects in the scene (many DCS
> nodes). I'm aware of the pfChannel man page caveat, "
>
> However, range sorting spans both transformation and draw callback
> boundaries. Thus a range-sorted scene graph with many transformations and
> expensive draw callbacks may suffer reduced performance due to an increased
> number of transformation and draw callback changes."
>
> Unfortunately, the singular matrices don't seem to go away when the view is
> reduced and the eat up an inordinate amount of CPU time (exception
> handling?). Fortunately the scene is still drawn properly. Is there any
> solution, short of "fewer DCS nodes in view"? GL_BLEND transparency and
> using the opaque and transparen bins is not an option since it's too
> expensive. Anyone hit this before?
>
> Greg Harrison
> Air Force Institute of Technology
>
> =======================================================================
> List Archives, FAQ, FTP: http://www.sgi.com/software/performer/
> Submissions: info-performer++at++sgi.com
> Admin. requests: info-performer-request++at++sgi.com

-- 
"Only the mediocre are always at their best." -- Jean Giraudoux 


For advanced 3D graphics Performer + OpenGL based examples and tutors:
http://www.dorbie.com/

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Wed Jan 06 1999 - 17:46:41 PST

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.