best pool water - reactor / realflow / particle flow?

[sv]

i'm really not satisfied with the normal way that i, and i think a lot of others in arch vis, do pool water - more or less a noise bump map on a box. I'm not talking about surface interfaces or caustics, because these are the easy bit!

I've tried to have a look into this myself before asking here, but haven't found out a great way to do this yet.

I want to do some sort of simulation for the water surface - something that will interact with the shape of the pool, things like stairs going into the water etc.

I've seen some great, great water before - joconnel and others did some sh*t hot reactive water in those t-rex milk commericals, that was way more than an animated noise map....

What would be the best way to simulate swimming pool water and movement? I've looked into realflow, but filling a pool can take some time...... at the moment, i only need a still mesh as the particular project i'm thinking about is only stills.
Has anyone had any experience with reactor - i know you can do water effects in this.....anyone ever used particle flow - i find it a pain in the arse when it re-builds it's mesh everytime it does an autosave, or a scene change, or anything at all....can lead to long periods of just watching your screen.

if anybody has got any good tutorials or links, or adivse, i'd really apppreciate it, and i'm sure others would if we could create a really nice pool water model thread for us all to share.....!

[Dynedain]

I would still use the animated noise bumpmap technique.. but here's how I would enhance it when wanting to deal with special events you mentioned (like railings, fountains, etc.)

Use something like Combustion or Motion to "paint" an animated noise map with the special events you need, then use that animated map sequence instead of the typical animated noise map. You'll get pretty impressive results, and it will work well as a displacement map as well.

The only time you need more is when you need to get reaaaaaally close to an object intersecting your water surface, and then you'd want to go with the fluid sim route. But for simply filling a pool, animated map are going to be your best bet.

In fact, that's how the big movie houses do it. They use animated geometry for the big movements (like waves). Sometimes they generate these using particle-based fluid sims, but even then they bake it to a geometry solution. Then a mixture of animated bump and displacement maps to get the fine detail, custom painting where needed. And finally particle systems to add things like spray and mist. For something like a pool I can't imagine a particle-based fluid sim like RealFlow to ever be practical. (Believe me, I've tried.)

[joconnell]

Yup indeed - the pool stuff on the milk ads were lowly dreamscape dynamics and a pflow setup for the splashes. I'd definitely avoid realflow for anything at all. I think one of the main things that lets it down is the meshing - it's got loads of lovely detail in the particle sim but when you mesh it it looks shit - one of the best things about flood and aquarius (fluid solver for tp - probably not going to be released publicly - http://www.magicpics.com/aquarius/aquarius_gallery.html) - their mesh solver is very similar to dimos vray iso and implicit surface stuff - you keep a hell of a lot more sharp detail in the sim without having any obvious holes or breaks in the continuity of the surface.

I'd say Go with good layered fractals and world space modifiers too - I'm going to be doing some huge wave shots soon along the lines of the statue of liberty in day after tomorrow and I'll be doing animated geometry and some particle hits in it...

[Jonferimonic]

'fluid solver for tp' - what is tp and why do you think software as sweet as this wont be released to me joe bloggs?
Ive played with RealFlow and its like learning russian.

[Dynedain]

'fluid solver for tp' - what is tp and why do you think software as sweet as this wont be released to me joe bloggs?
Ive played with RealFlow and its like learning russian.

Well, the reason why something like that won't be released to the public is because it's a wierd, custom-built tool that is very dependent on that studio's specific setup and workflow. Most 3D from TV and Movies relies on tools of this nature.

If you think RealFlow's interface is a pain to work with and has problems, you've seen nothing. Tools such as the one described would be even more difficult to install and learn as there is only internal documentation for them.

[Jonferimonic]

Ah ok, as mentioned about the big studios like Pixar who build their own custom renderers.
Wouldnt it be a profitable venture for the companies to create a branch to release a public version of their software and make some more money?

[joconnell]

Tp is thinking particles so I think you'd use the fluid solver as an operator and then there's a shader for final render than can render a water surface with a less blobby look than realflow - it's using a continuous surface rather than realflows metaball based surface.

On the page itself it mentions that it probably won't be released :/

[vlado]

Ah ok, as mentioned about the big studios like Pixar who build their own custom renderers.
Wouldnt it be a profitable venture for the companies to create a branch to release a public version of their software and make some more money?

Not really (well, at least not often). When you write in-house software, you only need it to work in a very specific situation - e.g. just for these 10 shots that you have in this film, for this very specific camera angle and for these given lights. If may crash or produce artifacts or look wrong if you change any of these parameters, or have a completely non-exstent GUI, but you don't care as you need just this very specific set of circumstances to work. A general-purpose tool on the other hand requires a lot of effort to make it useful under as many circumstances as possible, and there is also the technical support for it, someone must organize the sales, deal with clients etc... these may not be worth the effort in the end. Also, you may simply want to keep the tool for yourself (for example, because it will bring more work to your studio), rather than release it for everyone.

Best regards,
Vlado

[Dynedain]

Ah ok, as mentioned about the big studios like Pixar who build their own custom renderers.
Wouldnt it be a profitable venture for the companies to create a branch to release a public version of their software and make some more money?

Sometimes, yes. For example RenderMan is a Pixar product that was developed that way. But Pixar still has an insane amount of proprietary extensions and modules surrounding RMan that they don't release.

The biggest reason is support. Look how much work it is for Chaos to bundle up new builds, fix bugs, deal with licensing issues, etc. And that's not even considering all the compatibility issues and interface-design issues. Most of the bigger studios make money from their product, the CG that they produce. They aren't software development companies. The only reason why they have in-house programmers to create these tools is because the CG industry is a cutting-edge industry for computing. This means that a lot of times, there isn't a solution available off the shelf and you have to build it yourself.

Think of websites for example. That is also a relatively cutting-edge application of computers. There are applications out there that will give you a drop in place website, just like there are applications to make 3D modeling and rendering incredibly simple (like Sketchup). But if you want to do something really cool and advanced, you have two options. Pick up something already available and be satisfied with what it will let you do, or code your own solution. The same principle applies to CG tools.

[Jonferimonic]

Great thanks for enlightening me on that subject.
Sorry SV didnt mean to overshadow your thread.

[sv]

Sorry SV didnt mean to overshadow your thread.

that's ok, you could just redeem yourself by posting a really high res animated pool surface map.
(joke)


no worries - it's made for really interesting reading as i'm sure you'll agree....TP looks amazing, and yes, i hate using realflow as well

[Neilg]

Ive had good results by rendering out 2 normal maps of the hollermeish(sp?) water plugin at different frames, blending between them using a mix material and animating that with a UVmap moving slowly from one side to the other, then also using an animated bump with 'turbulence' type noise to give extra variation and get rid of tiling. Slight blue/green fog, all the usual water settings.

The normal maps need a different UV number so you can map them seperatley too.

I'm in a bit of a rush at the moment (just waiting for a test render) and ive not got it to a level i'm happy with yet, but I think with a bit of effort it'd work really well.

[Jonferimonic]

post some results hombre

[sv]

hollermeish(sp?) water plugin

sounds promising - any further info on this?
would love to see some samples if you have the time......

[Neilg]

Ive got a 500 frame animation, i'll have a look at getting it uploaded with the scene tomorrow.

http://charles.hollemeersch.net/ - here's the plugin anyway.