Maria Shugrina, a Senior Research Scientist & Manager at NVIDIA, has created a Gaussian Splatting Jupyter Notebook for people to try out.
For those eager to dive in, Maria hasn’t stopped at just creating the notebook. She’s gone a step further by providing a comprehensive tutorial, ensuring people can get the notebook up and running seamlessly. This guide delves into the required dependencies, including the Kaolin Library.
The Gaussian Splatting Jupyter Notebook is able to be modified and controlled virtually, allowing for people to try out Gaussian Splatting remotely. If there is interest in a repository of 3D Gaussian files for people to try out, please let me know. I have quite a lot of examples and can begin to compile them for people to experiment with. Barriers continue to fall and make these files accessible.
For those that would like to try it, check out her Github fork, here! It’s essential to note that this version stems from the original Gaussian Splatting GitHub, retaining the original licensing terms from Inria. While NVIDIA’s foray into Gaussian Splatting might seem relatively new, underestimating them, especially in the realm of radiance fields, would be a mistake. After all, they’ve been pivotal in propelling consumer radiance fields forward, especially with the introduction of Instant-NGP. With the radiance field sector burgeoning, NVIDIA is well-poised to harness its potential, likely leading the charge.
Maria’s name also might be familiar to you. She just won Real Time Live at SIGGRAPH this past August for her work on Text2Materials and is part of NVIDIA’s AI Lab in Toronto, run by Sanja Fidler.
