RetinaGS: Scalable Training for Dense Scene Rendering with Billion-Scale 3D Gaussians
This work addresses scalability issues in dense scene rendering for computer vision and graphics researchers, representing an incremental advance in training efficiency.
The paper tackles the challenge of training high-parameter 3D Gaussian splatting models on large-scale datasets by introducing RetinaGS, a model parallel training method that enables scaling to over one billion primitives and achieves state-of-the-art reconstruction quality.
In this work, we explore the possibility of training high-parameter 3D Gaussian splatting (3DGS) models on large-scale, high-resolution datasets. We design a general model parallel training method for 3DGS, named RetinaGS, which uses a proper rendering equation and can be applied to any scene and arbitrary distribution of Gaussian primitives. It enables us to explore the scaling behavior of 3DGS in terms of primitive numbers and training resolutions that were difficult to explore before and surpass previous state-of-the-art reconstruction quality. We observe a clear positive trend of increasing visual quality when increasing primitive numbers with our method. We also demonstrate the first attempt at training a 3DGS model with more than one billion primitives on the full MatrixCity dataset that attains a promising visual quality.