Image created with gemini-3.1-flash-image-preview with claude-opus-4.7. Image prompt: Using the provided reference image, keep the pure white landscape field, exact vertical type hierarchy, generous margins, and galaxy-punchout Milky Way starfield clipped inside all-caps letterforms, but replace ‘HEROES’ with ‘WORLD MODELS’ in the same bold condensed grotesque, replace ‘ALESSO’ with ‘DREAM SIMULATORS’ in the same light geometric all-caps, and replace ‘TOVE LO’ with ‘NEURAL PHYSICS’ in the same condensed grotesque, keeping ‘(we could be)’ and ‘FEATURING.’ unchanged with identical tracking, weights, and high-contrast galaxy texture.
Transformers are not the end game. AI still needs a breakthrough I talked to @FidlerSanja, VP of AI Research at NVIDIA, leading company’s Spatial Intelligence Lab, and she explains why ↓ And you should definitely watch our full conversation to understand where AI is heading
https://x.com/TheTuringPost/status/2046016440529248431
2 Weeks. New Tools. Infinite Worlds🚀 The World Jam is LIVE. Build the future of interactive 3D with Marble 1.1 + Spark LoD. Join our Discord to start building. More info below 👇
https://x.com/theworldlabs/status/2047373234174304473
Check out our effort on World Model that allows *ultra long* video generation! What is the secret? Check out the posts from @gene_ch0u below!
https://x.com/songyoupeng/status/2046780636145565828
Forget lidar. One single camera. Runs in real time & is open source: A streaming 3D model that reconstructs scenes live, at ~20 FPS, over long sequences. End-to-end. Optimization tricks, cleanup steps? Nope. And it beats both streaming and even some offline methods.
https://x.com/IlirAliu_/status/2045813408172315001
LagerNVS: Latent Geometry for Fully Neural Real-time Novel View Synthesis”” TL;DR: real-time novel view synthesis with implicit 3D-aware latent geometry, achieving SOTA quality without explicit scene reconstruction
https://x.com/Almorgand/status/2044777840835498111
Markerless tracking on every sword tip overlaid on live footage in real time. Fencing becomes a 4d geometry problem where both players are solving for the same point in space & time. Watching 3d trajectories render live like this is doing something to my brain I cannot explain.
https://x.com/bilawalsidhu/status/2046022001022640136
Neural Gabor Splatting: Enhanced Gaussian Splatting with Neural Gabor for High-frequency Surface Reconstruction”” TL;DR: augments Gaussian primitives with neural color variation and frequency-aware refinement to reconstruct sharp high-frequency textures with fewer primitives.
https://x.com/Almorgand/status/2046593635680915912
The Japanese take their digital twins very seriously. Some amazing LiDAR and CityGML 3D datasets released by them too.
https://x.com/bilawalsidhu/status/2046225662969561569
FlashDrive: Flash Vision-Language-Action Inference For Autonomous Driving – Z Lab
https://z-lab.ai/projects/flashdrive/
Holy crap, NVIDIA just made it drastically easier to create large scale explorable 3d worlds. No manual stitching of smaller 3d generations like other 3d models. Lyra 2.0 looks pretty damn impressive.
https://x.com/bilawalsidhu/status/2044681790195912972
Leave a Reply