[Sidefx-houdini-list] Point clouds
ben.schrijvers at guerrilla-games.com
Fri Apr 3 09:45:13 EDT 2009
I geuss you have to rebuild P to describe a plane in 3D that intersects
with your pointcloud.
It's what Francois already says.
If you just take the image X and Y pixel positions and use them as
inputs to a float_to_vector3 node you'll be able to sample the pointcloud.
But only in the plane near the orgin with a size of 1 x 1. So big chance
you have to transform the 0-1 from the pixels to something else.
I once made a COP to sample a pointcloud (particles) to get drips of
blood onto a floor.
Used a feedback loop in cops to make the blood spread and clump together.
It worked pretty well. But I had a grid in SOP's that I used to position
my COP's image space.
I used the grid point positions of the 4 corners to rescale the image
Andy Nicholas wrote:
> Yep, that's exactly what I'm trying to do, and it's presented as an
> example called "pcComp" in the set of examples that Mario linked to
> yesterday. Unfortunately, that example doesn't work either.
> After a bit of tweaking, I did manage to get it compiling correctly by
> putting an inline VOP in front of the pcopen and inserting the VEX:
> vector P;
> That allowed it to compile, but it didn't seem to work beyond that. I just
> ended up with a flat color for the whole image, no matter what I tried to
> do with the coordinates. Like you said, this was taking into account that
> the point cloud was in 3D and the image is in 2D.
> I'd agree, it seems like a strange thing to do to bring the point cloud
> into COPs, but I have my reasons. I need to perform some operations on it
> in COPs and then trace it back out into SOPs.
> Maybe this worked in an earlier version of Houdini, and something's
> changed and broken the COP functionality of the pcopen VOP?
> Sidefx-houdini-list mailing list
> Sidefx-houdini-list at sidefx.com
More information about the Sidefx-houdini-list