2D Gaussian Splatting Viewer
Project Overview
As a personal project, I developed the 2D Gaussian Splatting Viewer, a first publicly available tool designed to enhance the accessibility and usability of the "2D Gaussian Splatting for Geometrically Accurate Radiance Fields" research. Recognizing the need for a user-friendly interface to explore and manipulate 2D Gaussian Splatting data, I built this viewer leveraging existing libraries and adding significant features to improve visualization, editing, and overall workflow.
This project addresses the gap in readily available tools for interacting with 2D Gaussian Splatting, making it easier for researchers and enthusiasts to delve into this exciting field. By providing intuitive controls and a range of functionalities, the viewer empowers users to gain deeper insights and experiment with 2D GS scenes effectively.
Key Contributions:
- Developed an intuitive and feature-rich viewer from the ground up, based on the 2D Gaussian Splatting.
- Integrated various rendering modes for comprehensive data visualization.
- Implemented editing and transformation tools to enable scene manipulation.
- Added training capabilities directly within the viewer for streamlined experimentation.
- Continuously improved performance and added new features based on user needs and research advancements.
Project Details
- Category: Web Application, 3D Visualization Tool
- Project URL: Viewer Github
- Skills Demonstrated: Python, Viser, CUDA, Gaussian Splatting, Rasterization
Key Features
Detailed Feature List
Beyond the highlighted features, the viewer includes:
- Comprehensive Render Types: RGB, Edge, Normal, View-Normal, Depth, Depth-to-Normal, Depth-Distortion, Curvature
- Disk Visualization: Option to visualize splats as disks for clarity.
- Edit & Save Splats: Interactive modification and saving of splat data.
- Mesh Export (Edit Tab): Export edited scenes as standard mesh formats.
- Render Path and Preview: Tools for generating and previewing camera paths for video rendering.