smoke trail on path

[dobin]

hi,

I've been asked several times about this effect https://vimeo.com/61015448

so here is the hip file with some notes inside.

trail_smoke_on_curve.hip

Also I have a question to houdini gurus here please:

Is it possible to align the containers along the path in DOPs? If I have the tangent and up vector of every cluster point how can I get the orientation of every point from SOP and align the corresponding container in DOP? I've tried Point Position DOP but can't get it to work.

Thanks!

[Yassine]

Hi Dobril,

thank you for sharing, it's a very nice set up. I'm really enjoying playing with it. I wanted to ask if you have ever experienced the smoke trail cut at the cluster volume boundaries. Something similar to what you can see in the image below.

I remember you mentioning the vel from curve as the unified field and I wonder if that could be eventually caused in the areas where there is no velocity.

Regarding your question I'm not sure if that is what you are looking for, but I think it could be worthy to have a look at what the Parent Fluid tool does. Here is just a basic set up. Hope it helps

parent_fluid_example.hip

[narbuckl]

Hi Dobril, you can orient volume containers with the position data on the smoke object under the initial data tab

-Nathan

[dobin]

hi guys,

I've got the scene from this thread:

http://www.sidefx.co...ight=clustering

so all credits for this setup go to Rob and SESI...I've just tried to develop it further and adapt it to work on a fast moving object at long distance and during my work I've always faced the same problem with cutting the density at the container's borders. In the initial scene they use average velocity in 'Sourse Volume DOP' to blend the velocity among all containers and this works for slow moving and fast dissipating smoke but for а fast moving smoke you will need a different way to equalize all fluid forces between containers. And since every container is simulating independently and limited by borders I was thinking to zero out all internal pyro forces, buoyancy, diffusion, cooling rate, etc...and use external forces which stay relatively equal during whole simulation as generating velocity from the curve tangents. For adding turbulence you may try 'Gas Turbulence DOP' with a SDF from the curve as a noise mask like this:

http://www.sidefx.co.../dyno/pyro_look

Try different cluster size for more overlapping containers and play with CFL Condition and substeps in pyro solver as well.

Nathan, thank you very much for the suggestion. I've tried 'Point Position DOP' to get the points position from SOP and use it as a position data for the pyro object but it does what the instance points do anyway- taking the position but not the rotation. I would like to orient containers on points normals the same way as Copy SOP does:

not oriented:

oriented:

This will make every container occupy less space which will lead to a much more efficient simulation and I wish I have this as an option in pyro object under instancing tab.

Thanks!

[dobin]

here is a new scene file with containers aligned on the path:

trail_smoke_on_curve_aligned.hip

Dobril

[etudenc]

here is a new scene file with containers aligned on the path:

trail_smoke_on_curve_aligned.hip

Dobril

Hi Dobril, thanks for sharing your file - very, very, helpful!

I hope you won't mind my asking - the sim you posted on Vimeo, did you actually sim that using clustering, and if so, did the sim ever stall?

I finally got Hqueue working and tried simming my own file across three machines, but they all appear to stall before finishing the second cluster, and there are quite a few more clusters to go.

If you have any tips I'm all ears!

[dobin]

Hi Phil,

I've never tried to distribute the sim on different machines actually but yes I've simulated this using clustering on one machine and of course it's getting slower with time and took me about 6-7 hours to sim but never stall.

[etudenc]

Hi Phil,

I've never tried to distribute the sim on different machines actually but yes I've simulated this using clustering on one machine and of course it's getting slower with time and took me about 6-7 hours to sim but never stall.

I think I've got it working now. It seems one *has* to use the cluster node and not the user-designated clustered curve you made (which I much prefer) for it to work properly.

[Skybar]

here is a new scene file with containers aligned on the path:

trail_smoke_on_curve_aligned.hip

Dobril

Thanks a lot for this Dobin! However, opening this in H13 makes the fluid boxes rotate really wierd. It works fine in 12.5, but it seems like the "ROT" VOPSOP doesn't calculate the correct rotations in 13. I don't know whats changed and I'm quite awful with math so I have no idea what is going on in there, I've tried fiddling with it but it doesn't really change anything. Any idea what is going on?

Edited November 1, 2013 by Skybar

[edward]

I think many trig type vex functions were switched to use radians instead of degrees.

[Skybar]

I think many trig type vex functions were switched to use radians instead of degrees.

Ah I see. If I remove the 'Radians to Degrees' VOP that is after the Trig Function VOP, I get the same numbers as in 12.5. Cool

[tanyakim]

Hello, I need a little help on recreating this scene. So instead of having containers that touch each others edges, I'm getting containers encapsulating the next. I end up with a really big first container and a small end container. Any suggestions what should I look into? I guessed it was GasResizeFluidDynamic or the getting more points but lessening the padding and adding more points returned same results.

 

 

I have attached the hip file I'm working on and also the image above.

 

Thank you and happy holidays!

trail_04_copy.hip

Edited December 30, 2013 by tanyakim

[ChargingImage]

 

so here is the hip file with some notes inside.

 trail_smoke_on_curve.hip 

Hallo!

 

As a newcomer to Houdini I beg: is ther ANY chance that someone will make a tutorial how and why it works? I understand some nodes, but as you perfectly now, one mistake in Houdini or wrong corelation = not working.

 

Just asking Thank you for sharing this. Never even thought that all this stuff is possible. You're like demi-god...

[wandersonp]

The better way to learn is dig into the files ChargingImage.

[Rio.Dominia]

Nice job.  And thanks for sharing.  I have done this once before too.

[WLVL]

Hi!

First of all thanks for all the precious tricks shared here! I'm trying to use this setup to make a meteor trail.

 

I have a small problem:

 

In source, there is a copy sop use to visualize how the fluid instanced containers should look like.

The orientation of the boxes and the container do not match. 

The boxes' Y and Z axis seem to be inverted. Why? How can i solve it?

 

Thank you so much for helping!!!

meteor_orient.hip

[MaxGyver]

Had the same problem.  I had to keep the Clamp to maximum size checkbox checked on the Gas Resize Fluid Dynamic DOP node.  Yeah, I kno, your post is 2 years old.

[cudarsjanis]

If anyone still interested you can find some more help with this cluster setup here: http://flooz-vfx.com/

On .hip with rabbit trail.

One I especially found helpful was sop solver that blends/transfers velocity between clusters.

It took ages to go through all scrips I found, understand what is happening there and apply to shots I need.

Proper tutorial about clustering smoke trails would have helped me a lot. I bet others as well.

 

[henrricLiu]

i open all the file from you ,find the smoke container  align the curve have a problem. i have a good way to solve the problem . this is my way to rot the container by matrix .

rot the smoke container _liuyonghe.hip