Re: pfDCSScaleXYZ

Jim Helman (jimh++at++surreal)
Thu, 14 Oct 93 14:45:29 -0700

• Next message: Jean-Francois Richard 2275: "One more question on parallel loading with Performer"
• Previous message: Renee Strong: "Mixing X and Performer"

Looks OK.

If you will be doing lots of these (and I do mean lots),
it might be more efficient to retain the desired
rotation, scale and translations separately and combine
them into the DCS matrix, thereby avoiding the pfSqrt in
pfNormalizeVec3. Rotations are comparatively expensive
as well because of the pfSinCos.

As a reminder: there are (at least) two issues with
non-uniform scales.

        1) They slow down the rendering of lighted
        geometry because normals must be renormalized
        in the graphics pipe.

        2) In Performer 1.0/1.1, bounding cylinders
        around segments in intersection traversals
        (pfSegsIsectNode, pfChanPick) will not be
        transformed properly.

rgds,

-jim helman

jimh++at++surreal.asd.sgi.com
415/390-1151

• Next message: Jean-Francois Richard 2275: "One more question on parallel loading with Performer"
• Previous message: Renee Strong: "Mixing X and Performer"

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:50:03 PDT

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