[Sidefx-houdini-list] getting a local space for a transforming object

Andrew D Lyons tstexture at gmail.com
Sun May 9 00:07:38 EDT 2010


Yep - I have made a suite of SOPs based on this approach already.
Unfortunately they require me to eyeball the model in order to find two
points that are colinear with the desired up vector and two points that are
co-linear with the desired N vector so I can define a basis from which to
pull the inverse transform. With some irregular models, finding those points
can be kind of time consuming and error prone. I also want to be able to put
the system in a foreach loop to process large numbers of models in
preparation for DOPs. For these reasons I've been looking into automatic
systems. An oriented bounding box based system was promising for certain
models, but is prone to flipping with many models - assuming you can work
out how to get around the continual topology changes when rotating the
object being bound - so I've just abandoned that approach. This new paper
looks promising. Because the company I work for is going to throw all my
tools away when it closes soon I'm not undertaking any major development
projects at this point and am trying to get SESI to do it :) - but while I'm
waiting, there are some SOPs in 10.5 that will allow me to cobble together a
quick otl that will get me part of the way towards an implementation of this

Also - someone told me there was an OBJ in 10.5 that does this out of the
box. Will check that on Monday...


On 9 May 2010 11:37, David Johnson <daveinengland at gmail.com> wrote:

> We have a SOP that will take the normal of either a point or a primitive
> and
> transform the whole model such that the point or primitive is at the local
> origin and the normal is oriented along one of the axis. Just a simple
> transform and a rotate VOP. It also puts the rotation into a Detail
> attribute for use at the object level. It doesn't do what the paper talks
> about (automatically finding what humans would consider "up"), but it's
> quite useful for setting up animation controls at the object level. I'll
> see
> if I can upload it on Monday to the forum.
> Dave
> On Thu, May 6, 2010 at 12:03 PM, Andrew D Lyons <tstexture at gmail.com>
> wrote:
> > Do you work in a large pipeline where you occasionally need to remodel an
> > animating (transforming - not deforming) cache in some non-transforming
> > local space? Imagine if there was a SOP that could put any moving object
> on
> > the origin and align it with some cardinal axis - and which stored the
> > original space for later repositioning? It's easy enough to put an object
> > on
> > the origin - but how do you align it to an axis or get it upright?
> >
> >
> >
> http://www.cs.ubc.ca/nest/imager/tr/2008/upright_orientation/siggraph08_upright_orientation.pdf
> >
> > With a SOP based on this paper we could perform all manner of modeling
> > operations on a previously moving shapes as they sit on the origin not
> > moving, and then afterwards use an extracted centroid with N and up
> vectors
> > with a copy SOP to put the modified model back in it's original animating
> > space.
> >
> > How handy would that be!?! Any takers for an RFE?
> >
> > Cheers
> >
> > --
> > =======================================
> > Andrew D Lyons | Digital Artist | http://www.tstex.com
> > =======================================
> > _______________________________________________
> > Sidefx-houdini-list mailing list
> > Sidefx-houdini-list at sidefx.com
> > https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list
> >
> _______________________________________________
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
> https://lists.sidefx.com:443/mailman/listinfo/sidefx-houdini-list

Andrew D Lyons | Digital Artist | http://www.tstex.com

More information about the Sidefx-houdini-list mailing list