Better preview systems

[cb LLC]

can we get better approximate preview indicators?

1. For example, Outgoing velocity preview for liquids. It would be nice if we can get approximate arrows of how far water will shoot out from a cylinder (water hose) before slowly falling to the ground depending on velocity strength, emitter size and grid res. (just 2-3 factors. Nothing more) Just a simple approximation. Currently it's just trying over and over without not a single instant visual feedback. maybe a spline that changes curvature depending on strength.

2 Or how many smoke particles will be born per second on one square meter of an emitter with such and such outgoing velocity, grid res and particles value.

Thanks

[Svetlin.Nikolov]

Hey,

Fluid simulations, especially FLIP liquds and voxel based simulations of fire/smoke that we use currently in Phoenix involve quite a bit more computations than simple particle dynamics like pflow for example.

With pflow the calculations are so few and so simple, it's not a problem to evaluate each particle from frame 0 to frame 100 for example, without the viewport starting to lag, unless there are a ton of particles. This is what Max does - each time you change the timeline frame, all particles are computed from frame 0 to your current timeline frame and for small particle systems it's not a big deal performance-wise.

Fluid simulations, however, need much more calculations in order to get that extra bit that makes particles such as liquid, and voxels such as smoke, to behave like fluids and not like simple solid fragments.

A specific thing about fluid simulations, that sets them apart from rendering, is that each step in time depends on the previous step. You can't know what's the fluid going to be like in frame 100, unless you already have frame 99, and so on.

In short, this means that you can't approximate what a fluid simulation is going to produce at some point in time without also running a fluid simulation. Why don't we just approximate it with a lower res simulation? Because the changing the resolution might produce a very similar result, but it also might produce a very, very different result, depending on a huge number of factors. This is a general problem of present-day computations fluid dynamics. I'm hoping we are going to resolve it at some point, but we also have to balance between solving other practical problems as well, so it's gonna need time.

Currently Phoenix requires cleverness by the artists when it comes to doing many iterations. For example, you know changing the resolution could produce very different result, but having the grid narrower at the same resolution would help you iterate many times faster, while also getting results that would hold up for a larger grid. It depends on your experience and your judgement of the scene you are working on.

Of course, it would be great if we could eliminate this factor completely and you wouldn't have to deal with this at all. I hope we could come up with a solution that would completely automate this process...

Cheers!