On this page


Groups are named collections of points or faces.

The Group surface node lets you sort points or faces in its input geometry into groups based on an various criteria. Other operators may also create groups as a side effect.

Most nodes that operate on points or primitives give you the option to apply the node’s effects only to one or more groups in the input stream, instead of every point/primitive. So, you can identify groups of points based on certain properties, and then only apply some nodes to those groups, and not other points.

When a point is deleted, Houdini automatically removes the point from all the point groups it belongs to.

Ordered and unordered groups

A point group can be ordered or unordered. Unordered groups store their points in creation order; ordered groups store points in selection order.

When you select points in the viewer pane, a single click performs an ordered selection, while a marquee box generates an unordered group. (The only time bulk selections generate or maintain an ordered selection is when only one point is caught in the marquee box.)

To...Do this

Create a point or primitive group

  1. Set the selection type to points (choose Select ▸ Selection Type ▸ Points or press 2) or primitives (choose Select ▸ Selection Type ▸ Primitives or press 4).

  2. In the viewer, use the tab menu to choose Group Geometry.

  3. Select the points/primitives you want in the new group, then click RMB to finish the selection.

  4. In the parameter editor, set the name of the new group in the Group name field on the Create tab.

Use existing groups when selecting points/primitives for a new group

  1. Choose the Group operator.

  2. In the operation controls toolbar , choose the name of an existing group from the pop-up menu.

  3. Use the buttons in the toolbar to change the current selection based on the group selected in the pop-up menu:

Combine and filter groups

The controls on the Group node's Combine tab let you create a group that is a combination of existing groups.

  1. Create a Group node and connect the geometry containing the groups to it.

  2. In the parameter editor, click the Combine tab.

  3. Enter the name of the new group to create in the left-hand field on the first row.

  4. Enter the name of the group from which to take particles in the right-hand field (or click the box beside the equal sign so the exclamation point is visible if you want to start with particles not in the right-hand group).

    Once you set up the top row, additional rows become available to further combine and/or filter the contents of the new group.

  5. For the additional rows, on the right-hand side, enter the name of another group of particles you want to combine with or filter from the new group.

  6. Use the menu on the left side to choose how to combine/filter the groups.

  7. Use the menu to choose how to combine this group with the new group.

    Click the box beside the equal sign so the exclamation point is visible if you want to do the logical opposite of the operation in the menu.

Group syntax

Many geometry operators (SOPs) work on a specific subset of points or primitives. When you use the operator in the viewer, you select the points/primitives you want to use. In the parameter editor, this list of points/primitives is specified in the Group parameter using the following patterns.

You can use multiple patterns separated by spaces. The matches for each pattern are combined to form the group.




All points, primitives, vertices, or edges. To match no elements, use !*.


Match anything in a named group.


Component number n.


In SOPs other than the Group SOP, you can specify a named group of points/primitives (created with the Group SOP).

You can use pattern matching (*, ?, and [ ]) in the group name. For example, arm* includes all point/primitive groups whose names start with arm. ^ can be used in the pattern by enclosing the pattern in { }. For example, {arm* ^arm3*} includes all groups whose names start with arm, but not arm3.


Components numbered from n to m (inclusive).


Components numbered from n to m (inclusive) skipping every step. For example, 1-100:2 means every other number from 1 to 100.


Components numbered from n to m (inclusive). Use the first keep numbers and then skip every step after that.


Every component except the ones matching the pattern. For example, !1-10 means every point/primitive except the numbers 1 to 10.

pattern may be a numeric pattern, attribute pattern, or group name pattern.


Remove components matching the pattern from the results of the preceding pattern. For example, 0-100:2 ^10-20 means every other number from 1 to 100 except the numbers 10 to 20.

pattern may be a numeric pattern, attribute pattern, or group name pattern.


Components where attribute matches value, based on the op, which can be one of =, == (same as =), !=, >, <, >=, or <=.

For example, @v>0 will create a group of all points whose v attribute is greater than 0.

You can specify components using [index] after the attribute name, for example @Cd[2] will get the blue channel of the diffuse color (Cd) attribute.

As a convenience, you can also use .x/.y/.z, .u/.v/.w, or .r/.g/.b instead of [0], [1], and [2]. For example P.x to get the X-axis component of the position (P) attribute.

(If you don’t specify a component for a multi-component attribute type, the comparison will use the first component.)

For example, @P.y>0 will create a group of all points whose Y component is greater than 0.

For string attributes you must put quotation marks around the value if it contains spaces, for example @myattr="foo bar".

You can use wildcards (* and ?) in the value when using =, ==, and != on string attributes.

You can combine the attribute syntax and the range syntax:


You can also use the attribute syntax with an interger pattern, but you need to enclose the pattern in quotes:

@id="5 8 10 15"

@id="0-4 78"

You can use this to compare against intrinsic attributes as well:



Components with the given attribute containing one of the listed values. For example:


For string attributes, you must enclose values containing spaces in double quotes, for example:

@path=/foo/*,/bar,"some string"

Edge syntax

Some nodes deal with edges (the line connecting two points) rather than points or polygons. For example, the Dissolve node. In these nodes (or in a node that is set to a mode that affects edges), you can use the following syntax in a group field to select edges.




All edges in primitive number n.


Point number n / edges with point number n.


Edge number m of primitive n.


Edges between point number a and point number b.






Next steps

Guru level