Houdini 20.5 Nodes Geometry nodes

Attribute from Parameters geometry node

Creates a dictionary attribute filled with parameter values.

On this page
Since 18.5


Since Houdini 18.5, SOP geometry supports dictionary attributes. This node fills in a dictionary attribute on the detail (whole geometry) with parameter values from a certain node. You can use this to record the settings of a node so later operations can see how the geometry was constructed.

You can also use this node to convert node graphs into geometry, storing the individual node parameters and wiring as attributes.


The node always grabs the parameter values at the current time.



Choose whether to copy a specific node’s parameters into a dictionary attribute on the detail (whole geometry), or build a geometry representation of the nodes in a subnet.

Detail from Single Node

Creates a dictionary attribute on the detail containing a specified node’s parameters.

Points from Subnetwork

Outputs a set of points encoding the nodes in a node graph as points and their parameters/wiring as attributes on the points.

Points from Compiled Block

Like Points from Subnetwork, but the target node should be the end node of a compiled block. This node automatically encodes all the nodes that belong to the compiled block.

Point from Single Node

Outputs a single point with the specified node’s parameters baked in. This is useful when manually wiring a network.

The Points methods replace the input with the generated points (they do not pass through the geometry in the input).

Node Path

The node to read parameters from.

This creates a dependency on the target node’s parameters, but not on the data of the node itself. Editing a node’s parameter interface does not trigger dependencies, so creating or removing spare parameters on the target node will not recook this node.

Category Filter

The node type categories (for example, SOP, VOP) that can be encoded as points. This can be a space-separated list of names. The default is *, meaning all categories.


The name of the dictionary attribute to create and fill in with parameter data. When Method is set to Detail from Single Node, this is a detail attribute. Otherwise, this is a point attribute on the points created to represent nodes.

Parameter Filter

Only store the values of parameters whose internal names match these names/patterns. This can be a space-separated list of names and/or patterns. The default is *, meaning store all parameters. Note that multiparm instances are only included if they and the parent multiparm folders both match.

Flatten Multiparms

When turn off, the node stores a multiparm folder as a single item, where the value is a list of the multiparm instances. When turned on, the node stores multiparms as separate keys (for example, it would store a multiparm named value with three instances as separate value1, value2, and value3 keys). By default, this parameter is turned off.

Flatten Ramps

Similar to Flatten Multiparms, but for ramp parameters.

This parameter is only available if Flatten Multiparms is turned off.

Evaluate Parameters

Parameters that contain expressions are normally evaluated at the current time and the current values stored. However, it might not be possible or sensible to evaluate them outside the node’s cooking. Turning this off will store the values of constants but not evaluate expressions.

Include Locked Geometry

If the node is a locked SOP node, the geometry is packed and attached to the generated point.

Create Implicit Wires

Often there is implicit wiring in a graph. Spare inputs, block paths, and subnetwork outputs are all controlled by parameters or flags rather than actual inputs. Turning this on will cause polylines representing those implicit wires to be made explicit. Explicit wires better survive geometry operations as they are not sensitive to duplicate names or renaming of points.

Create Inputs and Outputs

When turned on, each input/output of the specified network or node will have an extra point created of type “null”. These new points will be added to the “inputs” and “outputs” point groups, which make it easier to compose multiple networks, as the inputs/outputs are distinct, simple, and well-defined. It also makes Invoke Graph easier to use, as the inputs and outputs are already registered.

Create Channel Primitives

When turned on, will create channel primitives from any animated parameters, adding them to the geometry. This only applies if Method is one of the Points methods, and Evaluate Parameters is turned off.

Update on Node Movement

The generated geometry records the location of all nodes, but this is irrelevant to how the network acts when invoked. Only when Update on Node Movement is turned on is the geometry recomputed when nodes move. This prevents unwanted recooks generated by node movement.

Update on Node Renaming

The generated geometry records the name of all nodes, but this is irrelevant to how the network acts when invoked. Only when Update on Node Renaming is turned on is the geometry recomputed when nodes are renamed. This prevents unwanted recooks generated by node renaming.

Geometry nodes