|
if you look at the documentation under sandy.core.transform.Transform3D:
/*===========================================
combineTransform
public function combineTransform(t:Transform3D):Void
This
method allows you to have a single transform3D to combine several
transformations. By combining some transformations, you can expect some
speed improvment since the matrix multiplication will be computed only
once. BE CAREFULL, this feature can be very hard to use correctly, so
you have to completely understand what you are doing. you must also
take under consideration that it is the first matrix which is
multiplied by the one in argument
/*===========================================
now
this suggests to me that if I have a few branches of transformGroups
before hitting my Object3D, the matrices for each transformGroup are
multiplied together at every render cycle. is that correct?
if that is the case,
is
there a way I can set the position of an object without using transform
group. e.g. if my scene is static and the only moving thing is the
camera. having to do matrix multiplication on every render cycle on
static objects would be unnecessarily expensive wouldn't it?
I
know little about 3D engine implementations and it is not my intention
to criticize the wonderful job Thomas aka Kiroukou has done. I'm just
curious if there's a workaround or if the observation is trivial in the
first place and wouldn't significantly affect the performance.
Powered by Invision Power Board (http://www.invisionboard.com)
© Invision Power Services (http://www.invisionpower.com)