PDA

View Full Version : Ice mesh deformed by PointCloud: Need Help!

 larikaram08-25-2010, 11:31 AMHi guys! I'm fairly new to Ice and i'm trying to deform a mesh with particles from a PointCloud (particles inside of the mesh) . With some research I have ended up building this ice tree! it works fine but there is a pop in the mesh points movement, its not fluid. I tried a linear interpolate, array average. fcurve nodes.. but no success!! how can i make the movement of the mesh more fluid like if the mesh was sliding on the particles . Hope soon to be able to share my work with the community! thx lari http://features.cgsociety.org/newgallerycrits/g04/342104/342104_1282733486_large.jpg
ThE_JacO
08-25-2010, 12:33 PM
There is an ICE specific sub-forum. Moving the thread there for you, next time please post in the right location for the subject :)

pooby
08-25-2010, 02:26 PM
What if you made a blend of the average shape over 5 frames. The current, 2 behind and 2 ahead.
Wouldnt that smooth it out somewhat ?

Dont ask me how to do that, but it must be possible.

ThE_JacO
08-26-2010, 01:43 AM
Just a guess as I've only glanced at the graph and am not entirely sure what you're trying to do.

The pop might be because when a particle gets close enough to trigger the get closest location node (you might have a distance threshold in there), the particle will still be distant from the surface.
if you want particles to affect the mesh only when they exceed its volume, then use a test for "point inside volume" connected to an if. If it returns true, you use self.pointpositions in the true slot, if it doesn't, it means it's left the volume and you want to snap to it, in which case you evaluate your current tree to the false slot.

Reducing such differences with averages and interpolation when things are so distant is less than ideal, as you might lose detail you actually want to preserve.

bear in mind the above will still jump around between points, based on the closest particle, if the cloud is dense and has a "thickness" to it when it expands.
That might or might not be what you're after.

larikaram
08-27-2010, 03:05 AM