On this page


You can create object node (OBJ) assets that are defined by a Python script instead of a subnetwork of nodes (File ▸ New operator type, click Python type, set Network type to “Object”). This example defines a Python object node that gets its transformation information from a file on disk.

This example shows how you can get raw transformation information, for example generated from another software package or from a hardware device, into Houdini using Python.

You can load the pre-made assets from $HFS/houdini/help/files/hom_cookbook/PythonObjects.hda

Cook implementation

  1. Open $HFS/houdini/help/files/hom_cookbook/xforms_from_disk.hip.

  2. Click Play.

    The xforms_from_disk1 object node loads its transforms from the file motion.csv in the same directory. When the object node cooks, it looks up the transformation matrix corresponding to the current time from the file, and sets its transform to that matrix.

  3. Right-click the xforms_from_disk1 node and choose Type properties to open the asset’s type properties window.

  • We created a parameter in the Parameters tab for the file name, just as we would for a normal asset.

  • The Code tab contains the Python code implementing the node’s logic.

# This code is called when instances of this object cook.

# Get the Node object representing this node
this = hou.pwd()

# Try to get the cached transforms from this node's cached data.
# We could be fancier here by checking if the file has changed
# since the cache was saved.
xforms = this.cachedUserData("diskxforms")
if not xforms:
    # Read and cache the transform matrices from the file,
    # using a function defined in this asset type's Python module.
    xforms = this.hdaModule().reload(this)

if xforms:
    # Get the transform for the current frame
    index = max(int(round(hou.frame())), 1) - 1
    if index < len(xforms):
        xform = xforms[index]
        # If the index is after the last transform in the file, hold the
        # final transform in the file
        xform = xforms[-1]

        # Construct a matrix object from the 16 floats, and set this object's
        # transform to the matrix

Reload function

You’ll often want to move functionality into helper functions/classes in the asset type’s Python module to keep the node’s actual code cleaner. In this case, we've put the function to reload the file in the asset module (on the Scripts tab).

def reload(this):
    Reloads the transformation matrices from the disk file, and
    caches them in the node's cached user data.

    # Get the name of the file from my parameters
    filename = this.evalParm("file")

    xforms = []
    with open(filename) as f:
        # The file is a simple CSV table of floats separated by commas,
        # 16 floats per line
        for line in f:
            # Strip any whitespace/newline from the line
            line = line.strip()
            # Split the string on commas to pull out the numeric substrings
            ns = line.split(", ")
            # Convert the strings to floating point numbers
            fs = tuple(float(n) for n in ns)
            assert len(fs) == 16
            # Add it to the list of matrices

    # Cache the transforms
    this.setCachedUserData("diskxforms", xforms)

The asset’s parameter interface has a Reload button. The callback script on the button simply calls the helper function in the asset’s Python module:


HOM Cookbook