Using NumPy inside Blender is often presented as a simple performance win, but in practice it comes with its own constraints, pitfalls, and undocumented best practices. This talk focuses on Blender-specific techniques for working with NumPy effectively, drawn from real tool and workflow development rather than generic examples. Unless you know how to use NumPy, it makes your code slower.

This workshop covers optimal methods for moving data between Blender and NumPy, reusable approaches for common geometric operations such as dot products and vector projections, and techniques for generating and visualizing data such as forces and vector fields directly in the viewport. It also highlights NumPy APIs and usage patterns that perform poorly in Blender’s scripting environment and how to avoid them.