LiveSplat: Real-Time Gaussian Splatting for RGBD Camera Streams Now Available for Public Testing

LiveSplat is an innovative algorithm designed for real-time Gaussian splatting, utilizing RGBD camera streams. To see what LiveSplat can do, check out the demo video provided below. LiveSplat_Alexa.mp4 Message from the Author LiveSplat originated as a component of a broader proprietary VR telerobotics system. After posting a video of its Gaussian splatting capabilities on Reddit, there was significant interest from users who wanted to experiment with it. As a result, I decided to make this component publicly available as LiveSplat. Please note that LiveSplat is currently in alpha quality. Due to limited testing resources, it may not run smoothly on all systems. If you encounter issues, and your machine meets the specified requirements, feel free to reach out to me for assistance. Additionally, while LiveSplat remains closed source for now, I am open to business opportunities. If you represent a company interested in licensing or integrating this technology, please contact me at mark@axby.cc. I hope you enjoy using LiveSplat! — Mark Liu Requirements Installation LiveSplat is not open source and is protected under a specific license. Ubuntu bash pip install https://livesplat.s3.us-east-2.amazonaws.com/livesplat-0.1.0-cp312-cp312-manylinux_x86_64.whl Windows bash pip install https://livesplat.s3.us-east-2.amazonaws.com/livesplat-0.1.0-cp312-cp312-win_amd64.whl Running LiveSplat To start using LiveSplat, you will need to create an integration script that feeds your RGBD streams into the LiveSplat viewer. The repository includes a sample integration script named livesplat_realsense.py, which is tailored for Intel Realsense devices. If you are using a different device, you will need to write a suitable script or modify the existing one accordingly. By providing this tool, the author aims to facilitate experimentation and encourage feedback from the community. Whether you are a researcher, developer, or enthusiast, LiveSplat offers a powerful method to visualize and work with RGBD camera streams in real-time.