To illustrate the workflow, lets take a quick look at a basic RBD simulation setup. There, a prefractured tower falls down on a ground and breaks apart.

• The SOP setup consists of an RBD Material Fracture SOP to break a hires object into pieces.The node generates points that determine, where the pieces are created. Alternatively, you can also connect already existing points to the node’s fourth input. Note that simulation points and fracture points are not the same!

• An optional RBD Configure SOP packs the input geometry and adds the relevant attributes to the simulation geometry (see the note below).

• The RBD Bullet Solver SOP performs the simulation.

Attributes play a major role in conjunction with the RBD procedural, because they establish the connection between pieces and simulation points. To address the procedural, it’s necessary to perform a couple of conversions. Basically, everything is down to path and name attributes.

• In Solaris, the path is the so-called USD primitive path and determines where in the Solaris scene hierarchy primitives will finally appear. The path attribute has precedence, but if it isn’t provided, then name is used (and would typically contain a relative path which is appended through a Import Path Prefix on the SOP Import LOP).

  To minimize the number of USD prims we also want to have a single Mesh primitive for all of the fractured pieces. The solution is to assign the same path to the entire geometry.

• The name attribute is directly created by the RBD Material Fracture SOP and assigns a unique name to each piece, e.g. piece0, piece1, etc. This attribute is used by the Bullet solver to identify separate pieces, so we use the path attribute to configure that all of the polygons should be imported into the same USD Mesh prim.

It’s not possible to pass name and path to the procedural directly, because they're already consumed by the aforementioned SOP Import LOP. This node loads the required geometry to Solaris. To overcome this restriction, you introduce a piecename attribute that is derived from path and name. The new attribute also avoids that you have to make changes on the default settings of the SOP Import LOP or your own definitions of path and name to control the scene hierarchy.

• The piecename attribute must live on the pieces and the simulation points to specify which piece a certain point corresponds to.

• The piecename simulation point attribute must be a full path and you can assemble it from the already existing path and name attributes. The reason is that the RBD procedural can handle multiple transformations simultaneously. The new path can be seen as a unique identifier.

Let’s say path is /world/tower and name is piece0. In this case, the piecename attribute’s resulting two flavors are:

• Primitive → piece0

• Simulation point → /world/tower/piece0.

The table below shows how the various attributes are modified when they pass the network nodes.

	Node	Action
	Name	Manually creates the path primitive attribute.
	RBD Material Fracture	Automatically creates the name primitive attribute.
	Attribute Swap	Renames the name primitive attribute to piecename.
	Attribute Promote	Converts the path primitive attribute to a point attribute.
	Attribute Wrangle	Assembles a piecename point attribute from path and name.

Let’s start with the path primitive attribute. On the Name SOP, make the following changes:

Assuming that you work with the optional RBD Configure SOP, you have to make the path attribute accessible for the simulation process. On the Attributes ▸ Transfer Attributes parameter, change the default entry to v w path.

The piecename primitive attribute is derived from the RBD Material Fracture SOP’s own name attribute. To rename the attribute, you can use an Attribute Swap SOP. Here are the node’s settings.

For the simulation points, you also need a piecename attribute. As mentioned in the introduction above, the points' piecename is a full path assembled from the name and path attributes.

Right now, path is a primitive attribute, but you need a point attribute. For the conversion you can use an Attribute Promote SOP and change the following parameters:

You also have tell to the RBD Bullet Solver SOP that it should transfer the path to its fourth output with the simulation points. Open the solver’s Output tab and go to Attribute Transfer. Change the default entry to age w path.

To assemble the piecename attribute you typically use an Attribute Wrangle SOP. On the VEXpression field, enter the following line of VEX code:

s@piecename = s@path+"/"+s@name;

Caching is an optional step, where you write out the hi-res piece geometry and the simulation points to separate files. If you have heavy and slowly simulating scenes with complex fracturing, you should definitely think about fracturing. However, caching for the simulation points is necessary if you want to use the procedural’s File Cache ▸ Points Location mode.

You can use the File Cache SOP for this process.

Now that you have the required attributes, you can switch to Houdini’s stage level where you bring everything together. Let’s take a look at the network:

You start with a SOP Import LOP. This node was already mentioned in the Attributes chapter and it loads the pieces' rest state to Solaris. When you move the timeline slider, the objects should be motionless. The node also reads the path primitive attribute that comes with the pieces to establish the scene hierarchy. When you look at the Scene Graph Tree, you should see /world/tower. Additionally, the objects also carry the piecename primitive attribute.

Now, lets take a look at the procedural’s parameters. The descriptions in the table below explain the various settings.

Parameter	Value	Description
Primitive Path	/$OS	Here you define, where in the Scene Graph Tree the procedural appears. The default entry reads the node’s name (houdiniprocedural1) to create the hierarchy. You can change this path to your own liking.
RBD Primitives	/world/tower	This path must point to the pieces you've loaded through the SOP Import LOP.
Piece Attribute	piecename	Here you add the name of the attribute that exists on the pieces and the simulation points. The default entry corresponds with the attributes you've created in the SOP network.
Points Location	Primitive	The default entry reads the simulation points directly from the SOP network. This means that you have use the two associated Point parameters to a) specify where in the Scene Graph Tree the points appears and b) let the procedural know where to find the points. With this option, it can make sense to add a Cache LOP and buffer the simulation to improve playback speed. With File Cache, however, you define the path to the file sequence with the points.
Point Primitive	/rbd_points/$OS	Here you can determine the points' scene hierarchy. Again, the $OS variable takes the procedural’s name to complete the path. Change this entry to your requirements.
Points SOP	/obj/tower/POINTS	This is the SOP path to the particles. Typically, you create a Null SOP for better accessibility.

When Preview Procedural is turned on, you can already evaluate the simulation in the viewport. You can also complete your stage by adding lights, camera, and everything else you need.

The Scene Graph Tree is now clearly arranged.