>>>>> "S" == Stephen Chenney <[email protected]> writes:
>> The order in which transformations appear in the VRML file must be
>> respected.
S> Yes, agreed. But there are two directions in which to apply the sequence.
S> The question is which direction is right.
>> In my eyes, this is no special Inventor issue, just basics of Computer
>> Graphics Transformations.
S> The "basics" of transformations make no assumptions about the order of
S> application. The idea that transformations be applied from the top of the
S> stack to the bottom is merely one possible method for doing it. It is
S> in no way fundamental to the use of transformations in general. Hence my
S> original point: The order MUST be stated in the spec.
I apologize for my formulation. Of course, the more specific and
detailed the spec is (without any Inventor assumtions) the better.
S> The other thing missing from the spec is which way a matrix node is
S> applied. Is it multiplied on the left or on the right? Hence, does the
S> translation part go on the bottom row or the right column? The spec says
S> nothing about this, forcing people to waste their time finding out. And
S> don't tell me that "everyone does it that way" because not everybody does,
S> and even if they did it is no excuse for omiting it.
Again, this is almost hidden in the spec and should be pointed out in
the MatrixTransform node.
In the explanation of filed "SFMatrix" there is an example that makes
clear that the last *row* contains the translation.
(If you multiply left or right depends whether you are using row or
column vectors -- just to avoid further confusion.)
S> As a final note I would add that it would only take the addition of five
S> or so lines of text to fix these loopholes, so the issue of "spec bloat"
S> does not arise.
I agree.
S> Steve.
bye,
Michael