Over the summer I got to read a little about 3D Gaussian Splatting, which is an innovative method to render complex 3d scenes from a video. It is not necessarily a device, but I thought it was an interesting technique to share. It is faster to render and produces unprecedented level of detail compared to more traditional methods using Nerfs. Instead of using polygons to model the subject the system uses a mathematical method called structure-from-motion to find the differences between each photos (frames from the input video) to model using dots or gaussian points in 3d space. Due to the nature of Gaussian points, I think it adopts a brushstroke-like style in the renders.
Examples from Polycam.
