RunwayML Plugin for Grasshopper 3D

RunwayML is a platform for creators of all kinds to use machine learning tools in intuitive ways without any coding experience. RunwayML allows you to connect state-of-the-art machine learning models to third-party applications via open-source extensions, plugins, libraries, and add-ons.

Vahid Azizi is the Author of RunwayML plugin for Grasshopper 3D, which gives users direct access to RunwayML from Grasshopper 3D and makes it possible to exchange data between RunwayML and Grasshopper 3D.


In the following examples, you can see how this plugin handles data exchanging between Grasshopper3D and RunwayML.

Generating Environmental Sensitive BIM Model with Machine Learning Algorithms

The project was developed to integrate machine learning algorithms (i.e. GANs) in architectural design. Initially, 12,500 architectural designs with 9-floor height were analyzed in case of building performance with the aid of environmental computational design tools (ladybug tools), and 1,427 inputs were selected by their level of energy use intensity in Tehran. Subsequently, the synthesized images containing floors boundaries were generated for re-training of the predefined GAN model with the Runway ML in 6,000 steps. Following, output images from our GAN model were imported to the grasshopper 3D and reverse engineered to the primary architectural model. Besides, Rhino.Inside enabled the integration of GAN based architectural model into the BIM software such as Revit with adequate details. Eventually, we imported our BIM model to the Lumion to create VR-ready image visualizations.

Creating geometry by machine learning in Grasshopper3D with RunwayML

Five hundred geometries are created in Grasshopper3D. 9 sections were created from every geometry and a picture was generated from them. Every picture is an encoded version of each geometry. In RunwayML these pictures were trained by StyleGAN for 3000 steps. 100 pictures that were generated in RunwayML, used in Grasshopper3D, and decoded to 3D geometries.

Using SPADE-COCO a RunwayML model in Grasshopper

In this model colors of labels are taken from the SPADE-COCO model in RunwayML and used in Rhino3D’s layers. SPADE-COCO generates pictures and theses modification can be seen real-time in Rhino3D


How pix2pix model in RunwayML can be used in Grasshopper3D 

In Grasshopper3D a series of colored surfaces were generated. These colorful pictures used in RunwayML’s Pix2Pix model for generating real images. in addition by using MIDas model the depth of each image were detected. The output of RunwayML was used in Grasshopper3D to generate Mesh.

Using PoseNet in Grasshopper

Ramin designed an interactive panel in which you can change the color and pattern of its lighting with your smartphone. In this example by using PoseNet model in RunwayML with body movement colors and patterns of the Panel can be changed.