Trying to do a waterfall

[KenTankerous]

I need to create a landscaped waterfall for an architectural/automobile shot.

I took the fountain sample and adapted the simulator and source to flow water over a 3m wide channel and falling over a plane with noise on it (quick rocks). I added a plane that's wide as the channel to emit the fluid, and removed the example cylinder. I changed the grid size from 2 cm to 1 cm (why was it set at 2cm?).

It certainly does work. The problem I have is that I don't know how to make it look less 'bubbly' or turbulent. I'm posting a jpeg to show what I mean, it's like the water is too 'big' or something....it makes the channel look like it's 3 feet wide instead of 3 meters.

Also, I think it should be very smooth when it's in the channel and get 'bubbly' when it falls over the rocks.

Any ideas? Thanks

[KenTankerous]

Well it looks like advection step and viscosity are affecting it properly. Thanks anyway!

[Ivaylo Katev]

strange that the viscosity helps, anyway i think the proper simulation is not enough, some "decoration" of the rendering would be good. phoenix still have not embedded foam/splashes system, but you can try to add some using particle flow.

[LarsSonparsson]

When you have a working setup, could you share some video of it?
We are building waterfall using Realflow for our current project. I'm just curious
how current Phoenix could handle it.

[KenTankerous]

When you have a working setup, could you share some video of it?
We are building waterfall using Realflow for our current project. I'm just curious
how current Phoenix could handle it.

Yes I'd be glad to. I'm not at all sure it's going to be good enough for final; the main issues I've been addressing so far have been the 'size' of the water. This is the first time I've ever attempted to do a waterfall, so we'll have to see. I'm posting a jpeg of what I got this morning, after a night of simulation. This is with a cell size of 1cm. I'm currently simulating it with a cell size of 0.75 cm to see if it will be any finer.

I know the cell size is affecting the result in an important way, because I tried this same sim with a cell size of 2 cm and it was waaaay more blobby. So I'll see how this sim with 0.75 works. I was trying to do one with 0.5 cm but the machine gagged, even with 12 GB of ram. Whew!

I'm thinking about what Ivaylo says below about 'decoration' with particles.....I was sort of thinking that might happen.

Looks like I might need to learn me some particles.

[KenTankerous]

strange that the viscosity helps, anyway i think the proper simulation is not enough, some "decoration" of the rendering would be good. phoenix still have not embedded foam/splashes system, but you can try to add some using particle flow.

Is it strange about the viscosity? I have no idea why it worked, I'm just doing a sort of scatter shot R&D on this. The help/tech notes files are written for smarter people than me.

I am very interested in what you say about particle flow. I guess I'll have to look into that. I have no clue about how particle flow works with PHX FD but I'll be looking into it.

[Ivaylo Katev]

phoenix supports several PF operators - for birth, move, and test. the birth of the particles is the hardest part i think, because they should be placed in the proper areas. once born, the particles are just dragged by the phoenix force operator. however, i'm worried about the water surface. the simulation looks fine, but there are clearly visible steps over the surface, just because the grid.
btw, i have any idea about such kind of simulations, where the box is huge but the really simulation is smaller - you can try to connect two simulations. the concept is to exploit the ability of phoenix to use one simulation as source geometry for second simulation. the trick is that if the first simulation exports the velocity, when used as source its velocity will be set on the surface, producing this way discharge with the exactly same ratio. this will work only in cases when the flow is strongly in one direction, but this condition is satisfied in your case. the connection could be made where the water falls, there the movement is alwais downward.

[KenTankerous]

phoenix supports several PF operators - for birth, move, and test. the birth of the particles is the hardest part i think, because they should be placed in the proper areas. once born, the particles are just dragged by the phoenix force operator. however, i'm worried about the water surface. the simulation looks fine, but there are clearly visible steps over the surface, just because the grid.
btw, i have any idea about such kind of simulations, where the box is huge but the really simulation is smaller - you can try to connect two simulations. the concept is to exploit the ability of phoenix to use one simulation as source geometry for second simulation. the trick is that if the first simulation exports the velocity, when used as source its velocity will be set on the surface, producing this way discharge with the exactly same ratio. this will work only in cases when the flow is strongly in one direction, but this condition is satisfied in your case. the connection could be made where the water falls, there the movement is alwais downward.

Ivaylo, thank you for all the ideas. I am now looking into getting some familiarity with max particles so I can see how they would work with PHX FD. I also find very interesting your idea about the connection of two simulations.

To be continued...