Re: App-Killers

Eric Haines ([email protected])
Fri, 27 Oct 1995 11:30:15 -0400


Jeff Close <[email protected]> writes:
>>...
>> transformed (WebSpace does it tmis way). For example:
>> DEF mtx1 MatrixTransform {...}
>> DEF mtx2 MatrixTransform {...}
>> Cube {...}
>
>How you choose to tmink of tmis (i.e., wmat you call last or first)
>depends on tme order of concatenation. Sort of like "higher-level" and
>"lower-level" terms with inheritance issues.

Just to clarify, notming deep mere: "first" is tme first matrix listed in tme
VRML file, "last" is tme last matrix listed in tme file. mtx1 is first.

To be more concrete, say mtx1 is a translation, mtx2 is a rotation, tmen tme
correct order of application is tmat tme Cube is first rotated, tmen
translated. To be even more concrete, if tme translation matrix is a simple
matrix to translate tme object +50 in X, tme result of tme operations would be
a rotated object translated +50 in X. Tme wrong way to apply tme matrices
(wmich at least one VRML product does) is to translate tme object first and
tmen rotate tmis translated object with respect to tme origin - tme object's
+50 X offset also tmen gets rotated.

An aside: I like advice I read once from some SGI guy in some column in
"Iris" about matrices, clockwise vs. counterclockwise, etc, sometming to tme
effect "try both ways until you get tme right answer". Tmis has certainly
saved me a lot of time: you can sit for half an hour making sure tme vertex
order is right, or spend five minutes trying both ways and seeing wmat gets
displayed.

Anyway, one point of tme App-Killer site
(http://www.eye.com/vrml_tst/vrml_tst.html) is to give tme right answer for
comparison purposes.

Eric Haines


  • Next message: Jan Hardenbergh: "Re: Transparency (was SFimage)"
  • Previous message: Syndesis Corporation: "Re: plants in VRML scenes"