RealFlow Official Forum

Hi

I´m still learning RF and go through the manuals. In the near future i will use "network rendering" in RF, based on 4 nodes. I´ve a question concerning IDOCs.
I think i´ve understand the basic concept, more or less.

I´m confused about the information in the manual "This means that RealFlow does not simulate interactions between the IDOCs, respectively the individual emitters."
I mean, often you have more than one emitter in a scene, and the particels of the different emitters interact, collide etc. among each other. In this case , its not possible to distribute the scene over the network ?

Sorry, for my basic question.

Best,
Thomas

If you really want your emitters to interact then it is not possible to use IDOCs. As the name already indicates, IDOCs are "Independent Domains Of Computation".

Thank you Thomas.

Let me ask in other direction. Is it possible to distribute a scene with two or more emitters and interaction between them over the network ?

Pinochio wrote:Thank you Thomas.

Let me ask in other direction. Is it possible to distribute a scene with two or more emitters and interaction between them over the network ?

No, this is not possible for several reasons. One of them is that it's very difficult to synchronize a fluid simulation different machines. Another reason is that the exchange of huge amounts of data over a network totally destroys the benefits from such a simulation.

Keep in mind liquids work is iterative work. You change a parameter (never change too many parameters per simulation run because you will lose track of what's going on) and find if this is of any help with the result you are trying to achieve, if it is pushing the simulation towards the right direction.

Having just one machine, you can just work on one version. When you are having more machines, you will be quicker because you can create more versions in the same time, making it easier to find out what works best for you.

FlorianK wrote:Keep in mind liquids work is iterative work. You change a parameter (never change too many parameters per simulation run because you will lose track of what's going on) and find if this is of any help with the result you are trying to achieve, if it is pushing the simulation towards the right direction.

Having just one machine, you can just work on one version. When you are having more machines, you will be quicker because you can create more versions in the same time, making it easier to find out what works best for you.

Helpful tipp, thanks Florian. I had the picture of network rendering in mind, like i do with maxwell. But i understand now, that the philosophy in this kind of simulation is different. I´m on the right way now, i assume.

Best,
Tom

This is a problem I am facing very often as a freelancer, coming to studios with no or not much experience in dynamic simulation. They just don't understand you could be working faster if there are more nodes because this doesn't actually accelerate the simulation process but the iteration process and so the development of the setup one needs for a specific task.