custom fields otl

[Benyee]

how can i find this tools in houdini,it mentioned in Peter Claes's reaserch paper "Controlling Fluid Simulations with Custom Fields in Houdini Master Thesis"

[pclaes]

This would be the obvious location... -> scene files and tools.

http://www.peterclaes.be/tutorials/PeterClaesThesis/index.html

[goofoo]

Hello Pclaes and other, please tell me where i'm can find source code from this research paper http://nccastaff.bournemouth.ac.uk/jmacey/MastersProjects/MSc07/MyaWin/mwin_thesis.pdf

He is wrote (p.5): All the source codes, HTML files of the source codes generated by doxygen, plugin, pdf documentation of this thesis, user guide in HTML format, and some rendered images and sequence are presented in the CD with the thesis. Big thanx.

[pclaes]

You could try emailing the author or Jon Macey (the programme leader of the MSc in Visual Effects at Bournemouth). That code is not publicly available as far as I know. Perhaps some of the current MSc students could help you with it too.

Also the hdk has changed quite a bit since then so some of that might not work (very well) with Houdini 10. (A lot of the information and papers are available though and you can probably have a look at the code written for blender or some of the cuda fluids examples.)

[Benyee]

Thanks Peter:)

[Benyee]

can you share the two color smoke emission hip file

[pclaes]

That's not in there? I'll have a look for it this evening.

[Benyee]

That's not in there? I'll have a look for it this evening.

i haven't found it yet,

i'm trying to build commercial version of the otls

[pclaes]

yeah, at the time I was in uni. I need to have some time to look into them again and perhaps upload them to the exchange. I think either they get converted automatically to commercial assets, or I could ask kindly .

[goldleaf]

Yeah, they get converted automagically. Very cool of SESI to set it up that way.

[pclaes]

can you share the two color smoke emission hip file

Hmm, the one with the two spheres is indeed not in there and I couldn't find that file, but it is pretty much the same dop network as the "color emission" file. Only in 3d and with a few more forces applied.

[Benyee]

what does the "Gas Build Relationship Mask" node do in the solver subnet,some parameters i didn't understand clearly,

and also the "index field" node,can someone give me some tutors,thank you!

[Benyee]

hi,peter,the advect_by_curve.hip file perhaps doesn't work well,there are some unused dop nodes,i can't understand some nodes' real function in the net,and i also red the corresponding help documentation,can you give some notes on some important nodes,thanks very much.

[pclaes]

I don't have the file in front of me right now. But which nodes confuse you most? From memory I remember there were a couple of nodes in there that were trying to contain the velocity.

You have to think of it similar to blendshapes or a compositing application. Basically what is going on is you try to build a custom optimal velocity field. With the GasCalculate you can blend and merge different fields together.

You can always attach a visualization node to the field to have a look at what the resulting field looks like.

So you have:

- A velocity field built from the tangent vectors of the curve. This will make your fluid try to flow along the curve, but it will fail as at some point fluid will spill out, it will go out of the turn and overshoot.

Therefore you need:

- The gradient from the density field. The gradient from a static density field based on the curve. This will give you a vector that points towards the "closest point on curve". So you blend that with the tangent to get a vector that "sucks fluid towards and along the curve".

The problem is that you need to make the decision to "add" velocity to the system of blend towards an average, or perform a combination of the two. If you add too much velocity it will go crazy and become unstable. If you blend too much it will look boring. The nice swirls you get from fluid are actually coming from a bit of the fluid "overshooting" the optimal field.

I hope this makes a bit more sense. If there are any specific nodes that you would want help on, just post the names of them here and I'm sure either myself or others will help you with them.

[ikarus]

i did a run through of your color field explanation in your paper, i got it to work fine but when i tried to get it to a renderable state i A. had to import the visualization not the actual field and B. only one of the colors would render, is there some workflow for rendering out visual custom fields like that?

[Benyee]

i appreciate your help very much,Peter

[pclaes]

i did a run through of your color field explanation in your paper, i got it to work fine but when i tried to get it to a renderable state i A. had to import the visualization not the actual field and B. only one of the colors would render, is there some workflow for rendering out visual custom fields like that?

The main thing to keep in mind is the "name" of the volume primitive. You can plug any field into the color if you want.

Some shaders will pick up the Cd, but not all. It is easy enough to add it in your shader though (have a look at the basic smoke to see how Cd is imported into the shader).

When you import your volumes they will have names, for mantra to pick up color you will need the fields Cd.x, Cd.y and Cd.z . You can use a name sop to establish these names. You can also delete specific fields you don't want based on their names.

You can just as well bring in any kind of custom field (you have advected from dops) into your shader so it will render. Perhaps you want to mix two colors, something like Cd.x Cd.y Cd.z mixed with glow.x glow.y glow.z . All you have to do is assure the correct names are mapped to the correct primitives.

What you cannot do is bring in glow.x glow.y glow.z and expect it to be picked up by the shader automatically as a color. You have to import and mix this manually. If you are using the basic smoke, it needs to be "Cd" - case sensitive - to be picked up by the shader.

Have a look at this very basic file to get the names associated with the volume primitives.

color_volumes_01.hip

[Benyee]

i confused by some nodes' function and the basic workflow about the solvers,i tried to build my smokeslover with the baisc dop nodes,of course include microsolver nodes.let's give a simple example,smoke from a ball.

i used the defaut smokesolver's inside stuff for reference,at the first simulation frame,what fields are zero,and what are non-zero,in my opinion,all fields are zero except the sourceindex field,right?begin simulating, the source field and velocity field and also the temperature field are evaluated by sourceindex with gas calculate node(it seems the source field is not),all these processes used the density field for dest blend factor,right?

but at this time,simulation frame 2 for example,where does the density field come from?who evaluated its value before？

[Benyee]

how to understand the relationship between parameter "Calculation" and "Blend Method" in gas calculate node?

when the node performs without or with blending,how can we get the result from this node respectively,is there any principle or formula?especially concern about data calculation order in the process.