The Big Y #99
Hi! 👋 Welcome to The Big Y!
Turning 2D images into 3D renderings is a complex and resource intensive task when using traditional methods such as laser scanning. Luckily, several companies are working to accelerate this task using Neural Radiance Fields (NeRFs) to reconstruct 3D scenes from 2D images taken at various angles.
NeRFs use natural light diffusion to create surface and volume representation in neural networks, which can then predict and fill in the blanks and create a complete 3D scene. Nvidia and Waymo have taken two approaches, one cutting down the rendering time and the other breaking down scenes to improve scalability.
Nvidia’s Instant NeRF takes a set of still photos along with information on the camera angles of the photos, trains a model in a few seconds and then renders the photos into a 3D space almost instantly.
Waymo has introduced Block-NeRF that decomposes a potential scene into individually trained NeRFs and combines the output into one large scene. The Block method can improve results from a smaller number of 2D images and introduces modularity, the ability to add sections of new scenes one at a time.
The increased digitization of many industries is leading to the use of Digital Twin technology in which virtual counterparts of real-world objects and environments can be explored. NeRFs can be used to more quickly build Digital Twin representations of large environments and accelerate training and testing of robots or autonomous vehicles.
AI-generated fake profiles have been popping up on LinkedIn as a new tool for companies to cast a wider sales net. These profiles are hard to detect and if you don’t know what details to look for, you might entirely miss the clues that distinguish the fake face from a real face.
Thanks for reading! Share this with a friend if you think they'd like it too. Have a great week! 😁
🎙 The Big Y Podcast: Listen on Spotify, Apple Podcasts, Stitcher, Substack