Jump to content

High Res FLIP Sim


Naeem

Recommended Posts

Hi,

I am working on a simple FLIP sim. When I simulate with Particle Separation set to 0.1 (11k particles at the start), it caches all 500 frames without any problems. However, when I lower the Particle Separation to anything lower, Houdini gets hung at some point. With Particle Separation set to 0.05 (88k particles), Houdini hung at frame 23 and with 0.02 (1.4m particles) it got hung at frame 19.

The scene doesn't really contain anything special, just a few static geos for collision. Any idea what could be causing the issue?

P.S. : Forgot to mention this is on a Quad Core machine with 8G of RAM.

STORM_GEO_V03_R03.hipnc

Edited by Naeem
Link to comment
Share on other sites

Which build are you using?

From the journals:

**************************

Tuesday, April 17, 2012

Houdini 12.0.599: Fixed a bug that crashed FLIP sims when the underlying grid exceeded 4 billion voxels.

**************************

I think separation and voxelres might be linked with flip, so lowering the separation, will increase you voxel count. That said, 4 billion is a very very high voxel count.

Might be something else.

Link to comment
Share on other sites

Which build are you using?

From the journals:

**************************

Tuesday, April 17, 2012

Houdini 12.0.599: Fixed a bug that crashed FLIP sims when the underlying grid exceeded 4 billion voxels.

**************************

I think separation and voxelres might be linked with flip, so lowering the separation, will increase you voxel count. That said, 4 billion is a very very high voxel count.

Might be something else.

Hi Peter,

I am using 12.0.581. The maximum I tried to go to is 1.4M particles. Also, I didn't change the Grid Scale or the Particle Radius Scale. So, I am not really sure if that's the bug causing the issue.

Link to comment
Share on other sites

Well it seems pretty consistent with running out of RAM. As you lower the Particle Seperation you're going to get more particles, which is crashing your scene quicker.

On the DopNet try setting it's memory limit and outputting an explicit cache

Link to comment
Share on other sites

Well it seems pretty consistent with running out of RAM. As you lower the Particle Seperation you're going to get more particles, which is crashing your scene quicker.

On the DopNet try setting it's memory limit and outputting an explicit cache

In my original post I forgot to mention that I am caching to disk and not simulating to RAM.

Link to comment
Share on other sites

FIXED: For some reason the Wind force was causing a problem. When I tried caching to disk after disabling it, it worked perfectly.

Most of the DOP forces use a particle's mass to determine acceleration due to the force. Unfortunately with the current FLIP object each particles' mass is a function of the particle radius, so as you decrease particle separation (and radius), you decrease mass, making the same force parameters more powerful.

Since you're also adding noise, occasionally you're getting very large velocity spikes, which is making GasAdvect take a long time to solve, since it's limited in the distance it can move a particle per iteration. If you turn on the Performance Monitor you'll also see that GasExternalForces is taking a really long time to solve.

So I would look at other options for a wind-type force. Probably the most flexible would be to add noisy velocity directly to the vel field with a GasFieldVOP plugged into the Volume Velocity input on the FLIP Solver.

I'd also use the Move Particle Outside particle collision method since you have static geometry, and it will be faster.

(Oh, and ideally please post to either odForce or SESI forums, not both. Just about everyone reads both anyway, and that way any responses will be nicely organized in one place for other peoples' future reference.)

Edited by johner
  • Like 1
Link to comment
Share on other sites

Most of the DOP forces use a particle's mass to determine acceleration due to the force. Unfortunately with the current FLIP object each particles' mass is a function of the particle radius, so as you decrease particle separation (and radius), you decrease mass, making the same force parameters more powerful.

Since you're also adding noise, occasionally you're getting very large velocity spikes, which is making GasAdvect take a long time to solve, since it's limited in the distance it can move a particle per iteration. If you turn on the Performance Monitor you'll also see that GasExternalForces is taking a really long time to solve.

So I would look at other options for a wind-type force. Probably the most flexible would be to add noisy velocity directly to the vel field with a GasFieldVOP plugged into the Volume Velocity input on the FLIP Solver.

I'd also use the Move Particle Outside particle collision method since you have static geometry, and it will be faster.

Thanks a lot John! I'll look into it.

Link to comment
Share on other sites

(Oh, and ideally please post to either odForce or SESI forums, not both. Just about everyone reads both anyway, and that way any responses will be nicely organized in one place for other peoples' future reference.)

Sure will do :)

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...