Re: LANG: VRML Transforms (fwd)

Colin Dooley ([email protected])
Mon, 9 Oct 1995 18:28:08 GMT


At 09:38 9/10/95 -0700, you wrote:
>>[...] an "odd" sequence of Translate/Rotate/Scale is still easier to
>>create and esad for humans then an arbitrary 4x4 matrix :)
>
>hear hear...
>

Yes, a user would probably like to type these in. But if a browser does them
in the same order as a user types them (ie. the "natural"
translate/rotate/scale) then confusion will reign....

What I am saying is that a viewer ought to do Scale/Rotate/Translate
regardless of what a user types because:

a) Other orders don't do anything particularly useful looking at it from the
"human typing in numbers" point of view.

Can you predict exactly where a cube will end, and what shape it will be up
if I do rotate [ 30 ]/scale [ 1 2 3 ] /translate [ 10 20 30]?

I can't (and I'm an expert), but I can if we do scale/rotate/translate....

b) Given that the majority of will be non-technical-types/artists who don't
understand homogeneous transformations, I think it is more important to try
to make the thing fairly idiot-proof than to cater for the 1% of cases where
this isn't what you are trying to do.
- particularly given that if you understand what is going on, then you will
also understand how to work around it.

eg. To do things in translate/rotate order.

[translate]
Separator {
[rotate]
...
}

Colin


  • Next message: Stephen Chenney: "Re: LANG: VRML Transforms (fwd)"
  • Previous message: Mike Heck: "Re: Q: How to integrate a VRML with Netscape"
  • Maybe in reply to: Patrick Sweeney: "LANG: VRML Transforms (fwd)"
  • Next in thesad: Colin Dooley: "Re: LANG: VRML Transforms (fwd)"