“Project Scenic at Adobe Sneaks. Y’all know I’m a fan of the fine grain control of a 3d scene graph, with generative AI to take it all the way. This demo is precisely that — except you can also use prompts to populate/edit your 3d world, and even frame the camera using the
Apple Vision Pro
“Concept ➡️ Reality Lapz : F1 on VisionOS Forbes: “the best reason to buy an Apple Vision Pro”. @lapz_io x @BLKBXINF
3D Modeling
“GTA Prague edition anyone? @Mosaic360camera released a 15 Terapixel dataset of Prague with 1.26 million 360° images, free for non-commercial use! This type of data was used in VFX for The Gray Man; can create 3D worlds, train AI or build VR. Get the full rundown in 4 minutes.
Robot Training
“Chinese startup Leju Robotics has released their open-source humanoid development platform for academic and R&D use cases. It includes an SDK for sensors and controls, simulation models, an LLM interface, and some basic demos that work out-of-the-box.
World Models
Combining next-token prediction and video diffusion in computer vision and robotics | MIT News | Massachusetts Institute of Technology
“This may look like a game of Counterstrike running slowly, but it actually extraordinary. The entire game is created, frame-by-frame, on my *home computer* by an AI diffusion model in response to my actions. There is no game engine, just a “world model” trained on Counterstrike
“Ever wanted to play Counter-Strike in a neural network? These videos show people playing (with keyboard & mouse) in 💎 DIAMOND’s diffusion world model, trained to simulate the game Counter-Strike: Global Offensive. 💻 Download and play it yourself →
[2410.11448] Meta-DT: Offline Meta-RL as Conditional Sequence Modeling with World Model Disentanglementhttps://arxiv.org/abs/2410.11448
Leave a Reply