PDA

View Full Version : Splashing water



icon kid
08-05-2007, 11:50 AM
I'm working on a project for a client that requires "splashing water."
For my animation, sheets of a gel-like material (20 ft. by 8 ft. by 3 inches and weighing 425 lbs. each) will be ejected from shipping containers (through a narrow slot at the bottom) from a height of 70 feet and land in the water creating a big splash indeed. My question is four-fold:

1) Should I use Particles or Hypervoxels for this.
2) How many polygons or particles should I use for the area of water that will splash?
3) Since the initial splash will occur mainly around the perimeter of the rectangular shape, shouldn't the shape of the splash area be like a rectangle, but having a rectangular hole in the middle?
4) How do I make the secondary splash? That occurs when a heavy, solid object falls into water; it displaces the water in its path leaving a momentary "crater" which then collapses into itself creating another, but much smaller splash. Below is a link to a test animation -- all the elements are there except the splash. I'm using Lightwave 7.5.

I had a problem converting the anim to streaming media (.asf) so it will take a minute to load.

http://www.jeffs-icons.com/LW/GG/Falling_Sheets.avi

SplineGod
08-05-2007, 01:05 PM
1. To some degree, yes.
2. Enough to sell it.
3. Probably
4. Create a 3d grid of 2point polys. Apply Clothfx to that. Metalink your water object to the 3d grid. Heres an example:
http://www.3dtrainingonline.com/examples/splash.mov

icon kid
08-05-2007, 02:49 PM
That's a nice effect, SplineGod, epecially how the area around the impact gets churned up. The only thing wrong is that no splash is produced as the orange sphere punches a hole in the water. No splashing is seen until the crater begins to collapse into itself -- I guess it's just a matter of tweaking. Thanks for giving me a starting point, though.

RedBull
08-05-2007, 03:57 PM
I would look at Dstorms new LW PhysX Plugin demo, and use that as it would be vastly superior than LW's PFX.

But i would use Blenders CFD for the splash, as it's just easier to get the results,
and i could do a Blender setup and export in around 30minutes and rendered in LW.

It would take me much longer to tweak PFX in LW...

T-Light
08-05-2007, 04:21 PM
But the problem with creating liquid displacement in either blender or even realflow for the scale involved (20 ft. by 8 ft) is the time needed to make a correct resolution splash will be gigantic. Nothing wrong with experimenting with them though and see if you can approximate results you like.

If you can render your sea scape splash initially using SplineGods method then experiment with collision particle FX you might find a result you're happier with a little faster.

However, as Redbull mentioned, DStorms new plugin might just do the job you're after. If you go for it, please post back and tell us how you found it to work with, expected results etc.

RedBull
08-05-2007, 04:55 PM
But the problem with creating liquid displacement in either blender or even realflow for the scale involved (20 ft. by 8 ft) is the time needed to make a correct resolution splash will be gigantic. Nothing wrong with experimenting with them though and see if you can approximate results you like.

However, as Redbull mentioned, DStorms new plugin might just do the job you're after. If you go for it, please post back and tell us how you found it to work with, expected results etc.

Yeah i think the splash would take me miniutes to do in Blender...
And solving in Blender, is way faster than Realflow.

I would take the the sheet/gel object, and use that to fall into the fluids in Blender, and it would give pretty good results.... But I'm fairly versed in Blenders fluids...

You would not need actual scale as it appears in the Video that the water is virtually unseen, so only the splash is visible.

Effectively you could use the same splash for every object, and just scale the splash for different pieces...

You may also like to reference these tutorials on splashes and water drops..
may help the ideas flow.

http://www.3dtotal.com/team/tutorials/gerardo_drop/gerardo_drop_01.asp
http://www.3dtotal.com/team/Tutorials/waterdrop_continued/waterdrop_01.asp

Good luck..

cresshead
08-05-2007, 06:40 PM
i'd probably use the FREE realflow special edition that was on 3dworld mag a few months back....it's limited to 50k particles but shoud manage most small fluid tasks okay like this.

cresshead
08-05-2007, 07:47 PM
was in the christmas issue 2006 btw of 3dworld mag.

an exstudent of mine is doing some really neat stuff with it over here:-
http://z12.invisionfree.com/razor1/index.php?showtopic=1526&st=0

http://www.3dthunder.com/animation/waterstreet01.mov

T-Light
08-05-2007, 08:16 PM
Redbull -

Yeah i think the splash would take me miniutes to do in Blender...
Ooooh sarky :D

And solving in Blender, is way faster than Realflow.
Didn't know that though so apologies if I'm wrong but I'm thinking of the last time I played with blender fluids (9-12 months ago).

For instance, Cress's examples above show what can be done with the demo of realflow with a 50,000 particle limit. If the fluid is flowing into something small (say the size of a paint tray) then that kind of liquid resolution is fine.

So, if say it took 50,000 particles to do justice to liquid flowing around a paint tray (12 x 8 x 1 inches = 96 cubic inches) then it's going to take 36,000,000 particles just to cover the volume of icon kids collision fluid, (240x96x3 = 69120 cubic inches (which is 720 times greater than the paint tray)). Never mind the actual splash or the volume of particles needed for the enveloping sea itself.

It's the resolution of the fluid I was referring to rather than the calculation time for small fluid sims. For an idea of what I'm talking about look at Cress's example. It looks great. But if you placed a human being to the side of fluid it would look very globular and unrealistic. It needs a much higher resolution for the size sim that icon kids after.

Just my 2c. :)

SplineGod
08-05-2007, 11:42 PM
That's a nice effect, SplineGod, epecially how the area around the impact gets churned up. The only thing wrong is that no splash is produced as the orange sphere punches a hole in the water. No splashing is seen until the crater begins to collapse into itself -- I guess it's just a matter of tweaking. Thanks for giving me a starting point, though.

You have to treat it as two things when using LW:
1. The behavior of the water surface
2. Generating water droplets and so on.

In that test I did the water cavitates in a convincing way, The droplets are created when the crests of the wave strike a plane setup as a collision object. The water surface only emits particles where it collides with the collision plane. Additional emitters can be added for effect.
You can produce pretty convincing water and splashes this way.

Youre other option as others have mentioned is to try something in blender and import it into LW.

Mr Rid
08-06-2007, 04:49 AM
I have never needed RealWave in RF but was reminded of this old splash tutorial scene from RF2.5 that rendered in under 30 seconds, (sorry, dont know why the preview is flickering).
http://www.box.net/shared/static/u5ttebg30k.mov

If curious, a free and fully functional RF demo is available.

richdj
08-06-2007, 07:14 AM
It's the resolution of the fluid I was referring to rather than the calculation time for small fluid sims. For an idea of what I'm talking about look at Cress's example. It looks great. But if you placed a human being to the side of fluid it would look very globular and unrealistic. It needs a much higher resolution for the size sim that icon kids after.

Just my 2c. :)

The scene cress posted up has 31k particles, yes it's blobby in places, but I did it purely as a test for getting stuff from RF to LW as I'm entering the realflow comp. I need to refine the fluid, but thats just a matter of time...

The only thing with the 3D world version is it's non-comercial.

Rich

T-Light
08-06-2007, 08:16 AM
richdj -

The only thing with the 3D world version is it's non-comercial.
I know, don't you just LOVE the small print :D

By the way, I wasn't knocking your sim, these things take a lot of planning to look good and I like yours a lot. It was the whole water droplet v's particle size thing I was getting at earlier with regards to iconkid's relatively massive scale fluid blocks, I just used your sim as an example of scale. Sorry if that came across as a bash, it certainly wasn't meant to :)

richdj
08-06-2007, 12:04 PM
No problem, I didn't take it as a bashing.

icon kid
08-07-2007, 06:52 PM
Wow! -- What a lot of responses to my question. Thank you all.

I've been given nearly a year to complete this animation plus two others about different aspects of this same project. I plan on getting Lightwave's latest version (9.2) in the next couple of months to make certain that I'm compatible with the latest plug-ins. Maybe I should get an array of the plug-ins mentioned even though the features in some of them overlap -- like collision detection and certain types of deformation patterns. The first thing on my list to go with LW 9 is a physics engine. Many of my anims involve falling, sliding or colliding objects and I've always keyed them by hand which never comes out looking thoroughly natural aside from being tedious and time consuming. So it's:

General Physics:

"LW's PFX, "Dstorms PhysX,...vastly superior than LW's PFX. -- or Blenders CFD"

Deformation:

"Clothfx" "Realflow or Blender...faster than Realflow"

And one thing that no one mentioned,

Fluid Mechanics:
Namely, Waterfx.


The droplets are created when the crests of the wave strike a plane setup as a collision object. The water surface only emits particles where it collides with the collision plane. Additional emitters can be added for effect.

I'm not going to wait for LW 9 and all the plug-ins -- I want to know right now if I understand this technique correctly (with the help of a picture, below):

http://www.jeffs-icons.com/LW/GG/splash_diagram-1.jpg

So you're basically setting up a polygon that water will strike (in wave form) and bits of water will be emitted as a result, just like waves crashing into a peer, except we're making the barriers invisible so only the crashing water is seen. But I'm assuming that the term "collision plane" doesn't mean literally one polygon? -- correct? -- because it would take four planes to cover the sides of a rectangular object. For a sphere it would require a cylindrical barrier, or even a conical one -- so that some amount of water can fly over the top of it as it goes through the surface. Of course, I'm aware that I can also move or morph the barrier object into different shapes during the "wave crash" sequence to tune the results.

The funny thing about the three drawings at the top is that they show arrows of Force going in all directions of a circle -- the yellow ones being visible sprays of water and the black ones being contained and unseen. That's because while I was making the drawings I missed SplineGod's point that the droplets appear with the collision of a wave, not with the collision of a particle. And this shows that I was thinking more of explosions. When you bellyflop into the pool, you cause two things: waves and an explosion, but the first thing noticed is the explosion of water, which I think means that some particle or hypervoxel method needs to be set up for this animation (if not an appropriate plug-in) -- for a wide, flat, 425-pound object to fall into water from 70 feet. On the issue of the scale of the object I'm using, and the number particles in the "fluid blocks" I might have to use, I have no idea what's going to be practical until I test this on my system.

And, yes, I will post back with my finished animation when it's done even though it will be months from now.
In the mean time, I would appreciate any other comments, suggestions or links to relevant tutorials or sample animations.

icon kid
04-15-2008, 06:06 PM
Well, it's been nearly eight months but I finally finished the animation described above. I ended up using only hypervoxel emitters for the splashes, and procedural bump maps for the sea surface. Half of these past eight months were spent just on searching for the right tutorials that would well explain how to use hypervoxels; some experimenting with RealFlow and a lot of experimenting with Lightwave's ClothFX and collision effects.

In the first half of the animation you'll see a series of sixteen splashes (1 second apart). Each one requires seven emitters. Four of the emitters provide the splash and the other three give the "splatter." Why four emitters? -- because I wanted to achieve a differentiation in the sizes of the water particles that shoot or spray out from the impact. My first approach was to have all the voxels very large as they rise up from the water and all grow small together as they fall back to the surface. When I studied a dozen Utube videos of "Bellyflop" contests I saw that the impact causes particles of all different sizes to erupt at the same time. A tight area right around the perimeter of the falling object comes up like a solid sheet so I used the largest particles for that -- and big enough to overlap each other. They also don't rise very high in the explosion. That's the first tier. The fourth tier of splash emitters have the smallest particles; they also shoot the highest and, naturally, take the longest time to fall back the surface. The second and third tiers have sizes and shooting distances between the first and last. So, a single emitter object, as flexible as it is, cannot give you the control to have some particles to be small and shoot high and go more upward while having other particle be big, heavy and shoot low.

The three "splatter" emitters are configured to erupt at the "end of life" of the last three splash emitters. Then I basically just clone this whole thing, move the clone to a new position, set the timing of these events to happen one second later than the one I cloned it from -- and keep doing this until I have sixteen splashes. The second half of the animation has 27 splashes but without any "splatter" -- not necessary because those are seen at a distance of more than 200 feet, where the impression of rain-drop sized splattering wouldn't be worth the effort visually.

This animation is not photo-realistic but it served it's purpose; it's not for a Hollywood movie but just to demonstrate someone's idea.

Here's a link to it:
http://www.jeffs-icons.com/GG.asf

Thanks again for the help.