3DGS splats, or 3D Gaussian Splats, are a powerful and innovative technique for representing and rendering 3D scenes in real-time.
3DGS splats, or 3D Gaussian Splats, are a powerful and innovative technique for representing and rendering 3D scenes in real-time. They were first introduced in the seminal paper "3D Gaussian Splatting for Real-Time Radiance Field Rendering" by Dejan Azinović, et al., published in July 2023. Since then, the field has seen a rapid expansion of research and developmen
Each Gaussian function, or "splat," is characterized by its center position, covariance matrix, color, and transparency.
The key idea behind 3DGS is to use a set of 3D Gaussian functions to approximate the scene's geometry and appearance. Each Gaussian function, or "splat," is characterized by its center position, covariance matrix, color, and transparency. These splats can be efficiently rendered using a specialized rasterization technique that takes advantage of their compact representation.
One of the main advantages of 3DGS is that it allows for real-time rendering of photorealistic scenes, even when learned from small samples of images. This is achieved through a combination of techniques, such as interleaved optimization/density control of the splats and a fast visibility-aware rendering algorithm that supports anisotropic splatting.
3DGS has a wide range of applications, including virtual and augmented reality, gaming, and 3D modeling. Researchers are continuously exploring new ways to improve and extend the capabilities of 3DGS, such as incorporating it into production rendering pipelines like Vulkan, DirectX, and WebGPU.
In summary, 3DGS splats are a groundbreaking approach to representing and rendering 3D scenes in real-time, offering a promising avenue for future research and development in the field of computer graphics and artificial intelligence.
