RayGaussX: Accelerating Gaussian-Based Ray Marching for Real-Time and High-Quality Novel View Synthesis
This work addresses the problem of slow rendering in Gaussian-based methods for real-time novel view synthesis, representing an incremental improvement over RayGauss.
The paper tackled the computational inefficiency of RayGauss for novel view synthesis by introducing acceleration strategies like empty-space skipping and adaptive sampling, resulting in 5x to 12x faster training, 50x to 80x higher rendering speeds, and up to +0.56 dB PSNR improvement on real-world datasets.
RayGauss has achieved state-of-the-art rendering quality for novel-view synthesis on synthetic and indoor scenes by representing radiance and density fields with irregularly distributed elliptical basis functions, rendered via volume ray casting using a Bounding Volume Hierarchy (BVH). However, its computational cost prevents real-time rendering on real-world scenes. Our approach, RayGaussX, builds on RayGauss by introducing key contributions that accelerate both training and inference. Specifically, we incorporate volumetric rendering acceleration strategies such as empty-space skipping and adaptive sampling, enhance ray coherence, and introduce scale regularization to reduce false-positive intersections. Additionally, we propose a new densification criterion that improves density distribution in distant regions, leading to enhanced graphical quality on larger scenes. As a result, RayGaussX achieves 5x to 12x faster training and 50x to 80x higher rendering speeds (FPS) on real-world datasets while improving visual quality by up to +0.56 dB in PSNR. Project page with videos and code: https://raygaussx.github.io/.