I'm trying to change an attribute on some rbd's (let's say color) based on their velocity using a sop solver.

it all works fine, except that as soon i attach the sop solver using a multisolver, the SDF is recalculated on every frame (probably houdini assumes that the shape of the geometry is changed?)

this is the same behaviour as the first example in the sop solver's helpcard, except that for my shot it's not necessary, as the shape of the object is not changed by the sop solver.

the result is that the collisions with a groundplane don't work anymore, and the sim is also much slower

any ideas on how to surpress this recalculation?

Under the Collisions::Volume tab of the RBD Object you can set a file to save the .sdf to. Set the operation to Automatic will write out the file when you first cook it and then use that file for future frames rather than recooking.

thanks for the tip, i tried this but it seems like the calculations are still done on every frame

this is not the normal behaviour? maybe something wrong with my setup?


i was gonna create a primitive attribute for the velocity of each fragment of a breaking geometry, my workaround is to extract this information after the simulation now with a dop expression, but has i have to do this for each fragment using a delete/copy network it's pretty slow. (still faster than simulating with the sop solver active)

i think it would be much faster to incorporate this information using the sop solver, if it didn't force the SDF calculation.

another way could be to calculate the velocity per-point using a trail sop, but this is not ideal for my situation

Aearon
thanks for the tip, i tried this but it seems like the calculations are still done on every frame

this is not the normal behaviour? maybe something wrong with my setup?


i was gonna create a primitive attribute for the velocity of each fragment of a breaking geometry, my workaround is to extract this information after the simulation now with a dop expression, but has i have to do this for each fragment using a delete/copy network it's pretty slow. (still faster than simulating with the sop solver active)

i think it would be much faster to incorporate this information using the sop solver, if it didn't force the SDF calculation.

another way could be to calculate the velocity per-point using a trail sop, but this is not ideal for my situation

i have the same task and absolutely the same problem!