Digital 3D spaces have exploded in that past few years from AR/VR to notions of the *metaverse*. However, these virtual/augmented scenes need to be populated with 3D assets that look good! Lots and lots of them, in fact. I wanted to create exactly that, but with a little twist—what if said assets could respond to the observer? I turned to growing digital plants that respond to user behavior. Virtual spaces can feel all too impersonal and empty, and adding dynamic greenery puts some of that life back in.

When I first conceived this research project, I knew I wanted to use procedural techniques to generate the plant geometry. I brainstormed a list of parameters I wanted to parameterize: leaf shape, leaf size, leaf curvature, stem length, stem curvature, and colors. I designed the plant "function" off these inputs, using an old geometry kernel I coded up with geometric curve operations. Upon generation, the plants are displayed in the browser—three.js is sure powerful!

To add responsiveness, all I needed to do was map behavior metrics to different parameters of the plant function. Using procedural techniques sure paid off. The metrics I decided to use were cursor variance, user position variance, and time of day. These potentially reflect a variety of emotions, from relaxation to anxiety, and an environment that validates or conflicts these feelings can, perhaps, make digital spaces feel sentient and personal. Affective computing is pretty cool, n'est-ce pas?