Trying to optimize my simulation times for a smoke simulation. 2 questions.

Question 1

On the Smoke Object node there is an option for ‘Closed Boundaries’

Do these boundaries act more like walls or like windows, that is, if enabled, is the simulation INSIDE the boundaries in any way affected (like walls) or by setting boundaries are you simply optimizing your sim by clamping how far your fields will have any effect?


Question 2

I'm using a GasResizeFluidDynamic node to optimize my simulation(reduce sim times). Max bounds is set to 'Clamp to maximum and Initialization Static.

Am I correct in assuming that optimization is the main purpose for this node and are my settings optimal?

Appreciate any feedback. have spent a few weeks now deep diving thru tutorials and master classes on microsolvers but still lots to learn!

The most efficient smoke setup is slightly different. It lack the dynamic bounding box and updates to the fluid sourcing, so it efficiently uses the gpu.

Check it:
https://www.sidefx.com/docs/houdini/examples/nodes/dop/smokeobject/OpenCL.html [www.sidefx.com]

Strange that example uses the old sourcing method. Isn’t Pyro Source + Attribute Rasterize supposed to be the ‘new’ workflow?

With H18 will have to relearn everything once more

Speed is ultimatly affected by the desired fx, mediated by optimization opportunities.

H18 is the total rewrite of the pyro solver - new unit tests will have to be performed. The takeaway for SideFx should be to produce a range of examples of optimised -> controllable, as Houdini is now being picked up by a lot of people who just push buttons - i.e. Maya uses