Generative modeling

This sad yet calm tiger was a pure accident from another project.

I originally wanted to create an object that was clearly both glitchy and elegant at the same time, with the end result being this spinning thing.

As I kept tweaking the parameters, all of a sudden this tiger face (but upside down) came into being, so that became the new direction. I love you, low poly sad red tiger.


Then we have this slowly evolving thing below. The original shape is a sphere, with two separate Wave modifiers and two noise texture modifiers, all animated over time.

For the material I focused heavily on pushing subsurface scattering to the limit in Blender’s realtime engine, Eevee. The varying and moving depth of the subsurface scattering is generated from two noise textures that interact with each other over time.


I do so love abstract art. This image, a personal favorite, was by abusing Blender’s subsurface scattering. I used as low a sample value as possible (a total of three subsurface passes) and turned off blending of the passes. It gives a wonderful 2d feel, with barely a hint of actual 3 dimensional depth.


These two images are from a series, where each figure is generated from a single mesh with no more than 10 sides. Everything else is handled by modifiers: duplication, placement, scaling, rotation. Colors come from a single color noise map, with each face getting a randomly assigned color.