[comp.graphics] binding of 3-D transformation matrices

rick@hanauma (Richard Ottolini) (02/16/90)

The problem:
Drawing a scene composed of duplicated simple objects is slower than
coding each object separately.  For example, the drawing of earthquake
source icons in an earth volume.  The same object is duplicated, but
independently oriented and scale each instance.  I suspect the graphics
system is pushing and popping and recomputing 4x4 transformations all the
time.
A solution:
A suggested solution is to have "dynamic" and "pre-computed" transformations.
The "pre-computed" type would be computed during the construction of the
image from sub-objects and remain fixed fixed for the duration of the
image.  The "dynamic" kinds would be used for interactive change of overall
perspective and reorientation of re-orientable parts of the scene.
Question:
Do any conventional interactive 3-D systems make this distinction?
I use two that are only dynamic.