Simulation Nodes

This is detail of realization. You shouldn’t thinking about this. And solving of this kind of cases

implementation details of attributes too.


This approach

can have problem in case there is single value is still field but not explicit (See Geometry Nodes: Simulation nodes not work with sampling nodes if value is single).

Main goal of single value fields, as i can see, is just make more explicit operation on non-field things. Trying to optimize this…?
But the very essence of the question has a problem. Users should not attempt this. It is not up to the user to determine where a group of nodes should make a field, but where a sing value. It depends on how well the field system is optimized.

Case simulations have a different problem:
Fields cannot live long. They should always evaluate to an attribute or single value if the fields are single.
But if this usually happens implicitly, then in the simulation the user must determine this himself.

In my opinion, the best thing to do here would be to simply expect the sim zone to deduce the status better (field or single). And in the interface, if it is not a field (single), the field settings for simulation sockets item should not be displayed.
This is just misleading.

I haven’t tinkered much with simulation nodes yet, but I am curious. When I worked on the spray/foam image baking for the ocean modifier, I opted to map the Jacobian XYZ values to RGB (with an inverse option in case I mucked things up) and had thought it would work nicely with the then-planned new particle system to impart speed in each axis. I’m not sure whether the simulation nodes framework would work with this to allow for oceans to now emit spray-like effects. I’m not sure where to start in order to find out…

That data ends up as an attribute I think? Which is easily accessible from GN.