Using Instance on Points and Realize Instances in Blender Geometry Nodes

What will you learn in this tutorial?

In this guide, you’ll learn how the Instance on Points and Realize Instances nodes fit into a Blender Geometry Nodes workflow. The first node adds instances of geometry to selected points. The second converts those instances into real geometry data. This lets you use an efficient representation when duplicating the same model throughout a scene, and make instances individually editable only when necessary.

You can think of examples such as elements repeated along a facade, plants in a landscape design, or furniture duplicated in an interior. These are illustrative examples of architectural uses for the nodes, not specific projects or ready-made recipes. The key question is whether the duplicated elements will remain identical or need to be changed individually.

Requirements and node roles

The required software is Blender; the workflow uses Blender’s Geometry Nodes features. The Blender manual referenced here displays pages for the “latest” version, but does not specify a particular Blender version number. The sources also do not state whether an account is required. For that reason, no version or account requirements are asserted here.

The main inputs of the Instance on Points node are Points and Instance. Points provides the geometry containing the points that determine instance locations; Instance provides the geometry to be placed at those points. You can use Selection to control which points are used, Rotation to control the instances’ orientation, and Scale to control their size. The node’s Instances output carries the generated instances.

Instances create references rather than duplicating the underlying geometry data when the same geometry is used many times in a scene. Realize Instances converts these references into real geometry data. This makes it possible to work on instances individually, but can significantly hurt performance when there are many instances containing complex geometry.

Step by step: Adding instances to points

1. Set the point geometry. Connect geometry containing the points where you want the instances placed to the Points input of the Instance on Points node. According to the source, the node works with geometry types that have a Point domain, including meshes, point clouds, and curve control points. The positions of the points affect the transformation of each resulting instance.

2. Connect the geometry to use as an instance. Make the geometry you want to duplicate available to the Instance input. This input can contain a single real geometry or multiple instances. The node adds geometry from this input to each selected point. An “instance” here means a repeated reference to the same underlying geometry; it does not mean that every copy starts out as independent geometry data.

3. Select the points to use. The Selection input determines whether an instance is created at each point. A true value creates an instance; a false value skips that point. This lets you create instances only at selected locations instead of using all the point geometry. The guide pages do not describe how to generate the data needed for selection.

4. Adjust size and orientation. Scale determines the size of each created instance, while Rotation determines its orientation. You can provide an Euler rotation to the Rotation input. The official manual also notes that rotation outputs from nodes such as Distribute Points on Faces and Curve to Points can be used. To generate an Euler rotation from a direction vector, you can use the Align Rotation to Vector node. This aligns the instances with the chosen vector.

5. Keep the result as instances. The instances created by the node can be passed on to subsequent geometry operations using its Instances output. If the input geometry contains an id attribute, that attribute is copied to the resulting instances. If the source geometry already contains instances, the node can also create new instances on the points inside them. This behavior is called nested instancing; the transformations of the new instances are affected by both the node inputs and the transformations of the parent instances.

Selecting from an instance list and converting to real geometry

1. Select from the instance list if needed. If the Instance input contains multiple instances, you can enable Pick Instances to use a specific instance at each point instead of the entire list. Instance Index determines which instance is selected and is used only when Pick Instances is enabled. By default, the point’s ID attribute is used; if it is missing, the index value is used. Negative values or values outside the list’s range wrap around from the other end of the instance list. The manual notes that this option is intended to be used with the Collection Info node.

2. Add Realize Instances only when needed. Use the Realize Instances node if subsequent operations need to change each instance as separate geometry. This node converts the input instances into real geometry data. Before conversion, changes applied to the same geometry are expected to affect all instances in the same way; after conversion, each instance can be edited individually.

3. Control the conversion scope. The Selection input on Realize Instances determines which top-level instances are converted. Realize All converts all nested levels for every top-level instance and overrides the Depth value. If Realize All is not used, Depth determines how many nested levels are converted within each top-level instance. These settings help you choose the required scope without unnecessarily converting every instance into real geometry.

Common mistakes and things to watch for

One of the most common mistakes is converting instances into real geometry at every stage. Realize Instances lets you edit instances individually, but converting many complex instances can seriously reduce performance. Decide whether to convert based on whether the next operation actually requires independent geometry.

Another common source of confusion is treating Selection and Pick Instances as if they perform the same function. Selection determines which points receive instances. Pick Instances selects which instance to use from the instance list and works with Instance Index. Also, Instance Index is not used when Pick Instances is off.

Nested instancing can also lead to unexpected transformations. If the point geometry already contains instances, new instances may also be added to points inside those instances. In this case, the result depends not only on the Rotation and Scale inputs, but also on the transformations of the parent instances. If the result looks unexpected, check whether the point geometry contains instances and review the conversion scope.

Next steps: Architecture and visualization workflows

For architecture and interior design teams, this method makes it possible to use the same geometry in many locations without repeatedly duplicating the underlying data. You can consider instancing with Instance on Points for arrangements such as facade modules, repeated furniture, or landscape elements. Realize Instances may be appropriate when each element needs different geometry operations; the trade-off is the performance impact, especially with many complex instances.

In your next experiment, first build an arrangement without converting the instances, then consider using Realize Instances only for parts that need independent changes. Observing point selection, Rotation, and Scale values separately makes it easier to understand which input affects the result. Also check the geometry types used in the project file and whether nested instancing is involved. Since the sources do not provide details about hardware, licensing terms, or compatibility, no specific requirements in these areas should be assumed.

Sources and license

This tutorial is adapted into Turkish from the “Instance on Points Node” and “Realize Instances Node” pages in the Blender manual. Both sources are licensed under CC BY-SA 4.0; the content has been translated and adapted.