Learning fluids

[Ezz]

Hi all good Houdini-folks.

With the introduction af Houdini 12 it has become very fun to simulate fluids.

I have done some productionsimulations in Maya and have experienced the pain it

is to achieve a good result. So I have been avoiding fluids for a while.

But now I am keen to learn not just the basics, but try to get a bit futher down the

rabbithole. I have worked a bit with Naiad and enjoyed the ease of using custom attributes

Normals, Color etc. as fields to drive some simulationparameters.

I need to get the same understanding in Houdini, but I am overwhelmed a bit about the complexity

and different approaches here.

I have a lot of questions like: "how can I controle a fluids velocity with colorinformation and normals from an object" and

so on.

I really need some basic understanding.

Does anyone of you know a good start, tutorial or paper that can enlightening me?

Thank you very much.

Erik

[edward]

Maybe some of these links?

http://www.sidefx.com/index.php?option=com_content&task=view&id=1834&Itemid=132

http://www.sidefx.com/index.php?option=com_content&task=view&id=1422&Itemid=132

http://nccastaff.bournemouth.ac.uk/jmacey/MastersProjects/MSc09/Claes/

[Ezz]

Hi Edward.

Thank you very much for your answer.

Still I find it very very intimidating. I guess its the nature of Houdini to be open as posible, but it sure

is hard to get my head around all this.

I will hope for a good tutorial on CmiVfx or like to crack this open for me :-(

[Ezz]

Hi.

I just found this site.

http://www.apileofgrains.nl/category/tutorials/

Here is a lot of nice information. In fact Coen Klosters builds a custom fluidsolver from scratch. Its simple and easier to see

whats going on.

Im gonna dig into this :-)

Cheers

Erik

[Ezz]

So I have been doing some reseach on how Houdini fluids are working. At first I was intimidated by the complexity of the inbuild solvers.

But going throug Coens example helped me a lot. Thanks Coen :-)

I have a better understanding of the concept with fields etc.

I have reverse engineered Coen setup for better understanding and it has giving me a faily good result. But I still miss a lot of extra bling bling ;-)

Here is a quick render of my sim:

Also I will attach my setup. I would really like to see some people come with some changes and suggestion to improve the setup. This could be an awsome post :-)

Cheers

Erik

eriks_fluidsolver_build_v01.hipnc

[eetu]

Also I will attach my setup. I would really like to see some people come with some changes and suggestion to improve the setup. This could be an awsome post :-)

Oh, doing this is surely a wonderful learning experience!

(*scribbles into todo-list*)

Some comments

- in your source fields, confinement, divergence and pressure do not have their Data Fields set, I suppose they should be named accordingly? (also, pressure has a bad explicit sop path for division size)

- I think it would be easier to create the fields with a Gas Match Field DOP, so you would not need to bother with all the linking

- add_velocity has a bad explicit sop path

- looks like you're advecting velocity with itself twice? ("gasadvectfield1" and "velocity_advect")

- I used to do the same as you, put parameters on a master null and link to them. Creating a subnet and putting the controls on the subnet level is a better practice in case you want to Assetize your stuff.

[Ezz]

Hi Eetu - You are ace !

Thank you very much for your suggestions. I have fixed the things you mention, and it allready looks better.

I have a couple of questions about the naming of the created fields like "pressure","confinement" etc.

How can I know wich microsolvers looks for these fields ?

Well "density","temperature","vel" is obviuos!

Also I have used the Volumefromattrib-SOP. Is that a new SOP in 12?

In the examples from Coen and Peter Claes they use

their own custom digital asset to transfer point-attributes to either scalar-fields or vector-fields.

Cheers Erik

eriks_fluidsolver_build_v02.hipnc

Edited March 28, 2012 by Ezz