At Floored, we build our own technology, crafting just the right tool for each job. We love taking down difficult new problems with extensible, open-source solutions.
Our research team's mission is to simplify and automate tasks currently performed manually by our 3D modelers. The goal is to create a system by which we can quickly and forgivingly capture data about spaces, and then automatically and accurately reproduce 3D models of those spaces. In short, the research team's job is to teach computers how to see.
To do this, the team has created both software and hardware. We implement the latest research in computer vision and machine learning to automate complex tasks including plane recognition, point cloud alignment, mesh simplification, and image-based 3D reconstruction. We have also built our own open-source 3D scanner to help collect HDR photography and depth data.
Until we've created tools to do it for us, the best way to build 3D spaces is by hand. That's why half of Floored's staff are 3D artists. We use a combination of tools, and the latest techniques in game asset creation, to recreate floor plans and photographs as interactive 3D experiences. And because our models live on the web, not on a game console, our artists must be masters of economy, wringing the utmost photorealism out of the most efficient, lightweight data.
Floored's graphics team brings our data to life as realistically, interactively, and beautifully as possible. We've built our Javascript-based graphics engine from scratch. With a focus on physically-based, progressive rendering, our WebGL pipeline incorporates advanced techniques including global illumination, HDR lighting, glossy reflections, deferred shading, and screen space ambient occlusion, all in realtime. We also tackle tricky optimization problems inherent to both 3D and the web, balancing sixty times a second the equation between gorgeous graphics and smooth framerate.
Our applications team builds all of the infrastructure necessary to capture, store, edit, serve, and display our 3D data. We work in a JS environment, using platforms like Node, Angular, and D3 to build our own tools. So far, these include a web-based catalog for our thousands of 3D assets, an online materials editor, our WebGL engine, and even an app that helps us pick what to eat for lunch. We also work in image processing, geometry optimization, and data compression to make sure our spaces are as responsive as they are pretty.