- Neuralangelo: stunning high-fidelity 3D surface reconstruction.
https://research.nvidia.com/labs/dir/neuralangelo/ - Magic3D: text-to-3D is coming fast!
https://research.nvidia.com/labs/dir/magic3d/ - Hair Simulation: efficient simulation of Discrete Elastic Rods (DER) for hair. This isn't quite an AI work, but very visually appealing.
https://research.nvidia.com/.../2023-08_interactive-hair... - Eureka: GPT-4 teaches a 5-finger robot hand how to spin pens better than I do! This is my team's work on bridging the gap between high-level reasoning and low-level dexterity.
https://eureka-research.github.io/ - Align Your Latents: high-resolution video synthesis with latent diffusion models. One of the best video generation works in 2023, coauthored by a few OG authors of Stable Diffusion.
https://research.nvidia.com/labs/toronto-ai/VideoLDM/ - Text2Materials: text prompt to generate materials like brick or mosaic that are tileable and seamlessly replicated over a surface of any size.
https://blogs.nvidia.com/.../siggraph-research.../ - CALM: a method to train steerable virtual characters to perform acts in a physics simulation.
https://research.nvidia.com/labs/par/calm/ - Vid2Player3D: learning Tennis skills for virtual characters!
https://research.nvidia.com/labs/toronto-ai/vid2player3d/ - Flexicubes: efficient and high-quality mesh optimization.
https://research.nvidia.com/labs/toronto-ai/flexicubes/ - eDiff-I: Text-to-Image diffusion with ensemble of expert denoisers.
https://research.nvidia.com/labs/dir/eDiff-I/
... 으로 끝나는 interactive-hair... 와 siggraph-research... 링크가 잘못 연결되어있네요.
각각
https://research.nvidia.com/publication/…
https://blogs.nvidia.com/blog/…
입니다.