[Python] Re-execute class method at each frame

[Coxyca]

Hi there!

 

I've been scripting in Maya for some years now and I decided to give Houdini a try but I still have some issues understanding how cooking works.

 

I'm trying to implement my own cellular automata (a classical 3D Game of Life). The algorythm is pretty simple: my class first generates a virtual 3D grid which is made of livings cells (1) and dead cells (0). Then, at each iteration of a custom user input parameter, it analyses the current grid and does some work to modify it.

 

But I have a hard time figuring how to do this. How can I have my script memorize its current state at each frame? For now the node just cooks entirely at each iteration, which is obviously pointless. Is there a way to execute only one method of my class? I thought about using a callback on the iteration parameter but I can't understand how it works in Houdini (I just made a "New Operator Type" and wrote my script inside the "Code" tab).

 

Sorry if this sounds stupid, I'm still learning! And I couldn't find anything useful on Google..

 

Thanks in advance for your help!

[Atom]

Your script needs to fetch the data from a stored location rather than just running as a 1st run each frame.

You can store data as attributes on geometry or userData on the python node itself.

 

Here is an example of userData I recently created. I needed to examine all the nodes in an imported rig and remove any animation so I can control them directly from code. I don't want to do this every frame, however, so I need a way to determine if i have done this before.

 

node_path = r"/obj/my_subnet/"
node_root = hou.node('%s%s' % (node_path,"my_node_name"))
if node_root != None:
    #Check to see if we have done this before?
    n = node_root.userData("my_data_name")
    if n == None:
        # Do work.
        node_root.setUserData("my_data_name",str(1))           # Set our flag so we don't do this everytime.
    else:
        # n is now a string that contains the value fetched from the userData.
        print float(n)

Edited December 8, 2015 by Atom

[mestela]

Use a solver; I've done this in vex wrangle but it should work the same with a python sop.

 

A solver gives you access to the previous frame, which you need for an iterative problem like 'game of life'.

 

game_of_life_solver.hip

[Coxyca]

Thank you very much, I'll try both solutions!

Is it a proper way to store this kind of data using setUserData though? (just curiosity)

Edited December 9, 2015 by Coxyca

[Atom]

In my case I think it was proper, I only needed to run code once. The iterative case of life seems more appropriate for a solver, however.

 

I was only using setUserData because the object I was storing the information on, was a bone controller which was a null. Initially I wanted to store an attribute in the object but the null does not have any Geometry, which is where attributes are stored, so I fell back to userData. You can use attributes as a kind of global storage system too.

[Coxyca]

Thanks for your help guys.

I've been trying to play with the solver but it's still a little complicated for me to have something entirely opaque for the user (this is a class project and the teacher wants it to be super user-friendly).

 

I'll keep trying to use a solver but I was wondering about something: is it a good way to do all the stuff inside a Python Module created in the Scripts tab? So that I can modify the grid using callbacks?

Say I have 2 parameters: a grid initial state and an iteration parameter which is supposed to be animated. I could call a function to create the grid when the initial state parameter is modified and then use a callback on the iteration parameter to modify that grid?

 

Or I could keep the core algorythm inside the Code tab and fetch the grid from the callbacks if it's possible?