PDA

View Full Version : Flowing sand simulation - approaches?


Pufferfish
04-04-2006, 07:22 AM
Hi,

I need to make some serious sand simulations so I decided to ask here what would be the best approach for the following?

Imagine a hand rising from the dry sand. Sand rises with the hand but starts to slip between the fingers etc..flowing down.
The reason why I ask this here in PFlow forum is because I would like to be able to do it with 3dmax/pflow.

I have done some tests with cloth simulation and opacity masks with added pflow systems, but I find it hard to give any volume to the sand. If you can share any tips/ideas, I would appreciate it. Thanks.

Ok, back to solving this.. :)

=JS

Pufferfish
04-04-2006, 10:25 AM
I made quick sketch of the actual event I want to make. Sorry about the quality, I hope it shows the idea.

What I am struggling with is the volume of the sand. I don't want it to be just a thin layer of sand particles but more like piling up on top of the hand and flowing down from the piles :)

http://www.jsaarinen.com/concept/sandhand.jpg

Tak Tak
04-04-2006, 03:18 PM
I think with RealFlow should get some nice result... I just downloaded trial version so I'll let you know if I get something to show :)

Pufferfish
04-04-2006, 03:40 PM
I think with RealFlow should get some nice result... I just downloaded trial version so I'll let you know if I get something to show :)

Yep... im "afraid" it will be the best sollution.

urgaffel
04-04-2006, 04:45 PM
The only thing I can think of that doesn't involve Realflow would be to have a mesh that is shaped like the piles of sand and have it deform as the hand rises and the sand "drains" away, maybe something as simple as scaling it to be nearly flat and then have particles either emit off it and slide off or have a ton of particles pre-emitted and have them slide off. Might work?

Ls3D
04-04-2006, 06:42 PM
If you need a RealFlow3 simulation node I have a license. Just try to keep it under half a million particles.

-Shea

Pufferfish
04-05-2006, 10:12 AM
The only thing I can think of that doesn't involve Realflow would be to have a mesh that is shaped like the piles of sand and have it deform as the hand rises and the sand "drains" away, maybe something as simple as scaling it to be nearly flat and then have particles either emit off it and slide off or have a ton of particles pre-emitted and have them slide off. Might work?

Yep, I did get some results with cloth sim. When the hand rises it actually lifts up a cloth with animated sand texture and also works as emitter. I also modeled the sand piles on top of the hand and used that version of the model for the cloth simulation to make the cloth form those piles. This way I got some results, but it needs some serious tweaking, especially what comes to textures.

I quickly animated noise map which pushes pixels from the middle (top of the hand ) to the outter edges of the map.. like very dense starfield simulation. Mapped it to the clothes and used opacity masks to fade the clothes away when I wanted to stop the sand flowing. It worked kinda nicely but not really happy yet. :D

EDIT: Ls3D thank you.. but I will first try to manage with pure max/pflow etc.. :)

erilaz
04-05-2006, 11:27 PM
Pufferfish, have you tried using a variation of allan mckays drip tutorial?
http://www.allanmckay.com/html/newtutorials.htm

This is actually an interesting topic! I might have a go at my lunchbreak!

Pufferfish
04-06-2006, 05:31 AM
Pufferfish, have you tried using a variation of allan mckays drip tutorial?
http://www.allanmckay.com/html/newtutorials.htm

This is actually an interesting topic! I might have a go at my lunchbreak!

Hi Erilaz!

Actually yes, I bought the Allan's DVD some time ago and have to say its great. I will partially use similar build as in Allan's drip system, but with cloth sim etc.. I guess I have to build the scene from all kind of different tricks to make it work :) There are also other cool things in that DVD that might work very nicely. I did play with the "acid burn" procedural method already, but I would have to make specific animated textures to get it look at least somewhat right. Cool little trick but animated textures are evil :D

Give it a go if you got any spare time and nothing else to do :D

Cheers :beer:

Ls3D
04-07-2006, 12:10 AM
Looks like a RealFlow version would need some serious drag force to look like sand, that and more density, surface tension, attractor daemon, etc.. a few days research anyway. This preview (http://www.shea-design.com/realflow/shea_RF3-arm-previews-web.mov) is 278,300 particles or 26 Mb's per frame. The emmiter is the white outline and it's particles were first settled into a 10x10x2 inch pool with the LOCK feature and a speed daemon. The second short preview is from a Zero G test - crazy stuff.

So how is the PFlow work comming?

-Shea

Pufferfish
04-13-2006, 06:12 AM
Looks like a RealFlow version would need some serious drag force to look like sand, that and more density, surface tension, attractor daemon, etc.. a few days research anyway. This preview (http://www.shea-design.com/realflow/shea_RF3-arm-previews-web.mov) is 278,300 particles or 26 Mb's per frame. The emmiter is the white outline and it's particles were first settled into a 10x10x2 inch pool with the LOCK feature and a speed daemon. The second short preview is from a Zero G test - crazy stuff.

So how is the PFlow work comming?

-Shea

Cool stuff.. yep, particles in your test (first) looks a bit too much like water. Second one was crazy indeed :D
I have found a sollution at least partially :) I still think I have to build the scene from several different systems, but managed to do some decent tests. I will post example clips asap.

Scale of this scene is HUGE. It's massive hand that rises :) So its not all pure clean sand.. there will be larger rocks as well, but the sand part is what needs lots of work :D

BrandonD
04-13-2006, 07:15 AM
FYI, there's zero viscosity to dry sand, so a fluid sim may not be all that advantageous.

In my experience, long before we had fluid sim for this kind of stuff we used to fake it by animating geometry to mimic the look of a large chunk of sand sloshing away. From there we'd emit particles along the surface to create the look of sand bleeding off.

Banshee
04-27-2006, 02:35 PM
@Pufferfish: So thatīs been some days... have your attempts been successfull?

Iīd be really interested in how this could work with particleflow as Iīve recently tried something quite similar and couldnīt handle it with pf because of the incredible amount of particles that would have been neccessary... After Max took 3 GB of memory (3GB switch activated) the rendering setup was canceled.


So -would you like to show us your animation and give us some hints?

Thanks in advance.

Pufferfish
05-05-2006, 11:03 AM
Iīd be really interested in how this could work with particleflow as Iīve recently tried something quite similar and couldnīt handle it with pf because of the incredible amount of particles that would have been neccessary... After Max took 3 GB of memory (3GB switch activated) the rendering setup was canceled.

I also cannot do this with pure sand particle simulation. It would take way too much power. Therefore I'm trying to use partially sand particles and partially fake methods to build up believable scene. What comes to tips I don't have any particular tricks to tell at the moment, most of them are discussed here. Animated textures/clothsim/different particles (facing with textures etc.) :)

I will post example when I get it to look at least some what good :)

happychopper
05-12-2006, 08:58 AM
I know zilch about RealFlow, but have experience in the Glu3D fluid simulator.

Use a surface connected to the hand as a guide for Glu3D particles to trickle off the hand:
Create a spidery shape to represent the drip channels and project it down to conform to the top of the hands surface making sure the drip channels fall off down the sides and onto the surface you wish to drip onto. Pile a load of GLu3D particles onto the middle of this guide object and turn adherance up (so they partially stick to the surface, thus conforming to the guide obhect's drip channels).

Run the sim and my guess is after some (ha ha, alot of) tweaking you'll be getting a pile of particles draining off the guide object and down the sides. Problems will be what happens when they hit the ground (unless it's not in shot). One other problem is Glu3D particles tend to bunch up a bit so it may look a bit like slightly wet sand and not the powdery bone dy stuff you're probably lookin' for. Hmmm.

If I didn't have a deadline this afternoon I def try this one out! Good luck.

edit: Just read the previous post (can't e pure particles) - sorry for me tab in the dark!

vancliff
06-25-2006, 11:30 PM
I would like to contribute to this thread with this sample scene i did in max 8. Is about a dissolving rock using noise and slice deformer over a sphere. Hope you like it.

blinxpro
07-10-2006, 10:05 AM
Looks like a RealFlow version would need some serious drag force to look like sand, that and more density, surface tension, attractor daemon, etc.. a few days research anyway. This preview (http://www.shea-design.com/realflow/shea_RF3-arm-previews-web.mov) is 278,300 particles or 26 Mb's per frame. The emmiter is the white outline and it's particles were first settled into a 10x10x2 inch pool with the LOCK feature and a speed daemon. The second short preview is from a Zero G test - crazy stuff.

So how is the PFlow work comming?

-Shea

Have you done any more sand simulation tests? I have been trying a few. I am attempting to get a hourglass effect, but the pile of sand in the bottom of the glass doesn't quite get enough drag. The pile settles too much.

It is difficult because sand acts with high viscosity when being poured, but RF wants to treat that like water and not let it build up upon other particles.

Is there a way to set more control over particle to particle interaction?... ...collision tolerance for instance.

CGTalk Moderation
07-10-2006, 10:05 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.