Confining RBD simulation inside volume

[Doum]

I'm still at the beginning of learning houdini. I just builded a brickwall and did a basic simulation

I'm searching a way to limit the rbd to simulate only what I can see in the camera. I think I need way to check if a rbd object is still inside a geo or outside and use the active node to make the one outside inactive?

First do you think it will optimise the simulation time, if not Is there a better way to decrease simulation time by not simulating what we are not seeing in the camera...

Thanks in advance for your input.

Doum

Edited January 24, 2013 by Doum

[gui]

one option is to use the active value node and group objects by expressions. Just use and dopoption expression in the object position, and set a threshold for that. If I´m not mistaken, to get the object centroid you must add its pivot to its translation

[bloomendale]

Create volume from camera (volume SOP) and use it to group points inside. use all points of all rbd objects. if at least one point of the geometry data of the object is inside group - you should simulate this object.

[Doum]

Thanks, I'll give it a try today

[rafaelfs]

I can think of two issues with this method:

1) If the camera is moving, some object might be deactivated in the beginning of the shot so when they finally get framed by the camera and become active their behavior might be inaccurate.

2) Some objects that were outside the FOV might enter it as a result of a chain reaction, supposing all the objects are being simulated all the time. Again, if they are disabled for being out of the FOV they might never appear in camera.

To solve the initial problem I'd suggest a first pass using a SOP solver to detect if an object will ever be on camera and storing that as an attribute. Then use that attribute in a DOPs sim to activate the objects that will ever be framed in the shot all the time. Does it make sense?

Cheers