pixelSplat: 3D Gaussian Splats from Image Pairs for Scalable Generalizable 3D Reconstruction
This addresses scalable and generalizable 3D reconstruction for applications like novel view synthesis, offering real-time efficiency and editable outputs, though it builds incrementally on existing Gaussian splatting methods.
The paper tackles 3D reconstruction from image pairs by introducing pixelSplat, a feed-forward model that predicts 3D Gaussian splats for radiance fields, achieving state-of-the-art performance on RealEstate10k and ACID datasets with 2.5 orders of magnitude faster rendering.
We introduce pixelSplat, a feed-forward model that learns to reconstruct 3D radiance fields parameterized by 3D Gaussian primitives from pairs of images. Our model features real-time and memory-efficient rendering for scalable training as well as fast 3D reconstruction at inference time. To overcome local minima inherent to sparse and locally supported representations, we predict a dense probability distribution over 3D and sample Gaussian means from that probability distribution. We make this sampling operation differentiable via a reparameterization trick, allowing us to back-propagate gradients through the Gaussian splatting representation. We benchmark our method on wide-baseline novel view synthesis on the real-world RealEstate10k and ACID datasets, where we outperform state-of-the-art light field transformers and accelerate rendering by 2.5 orders of magnitude while reconstructing an interpretable and editable 3D radiance field.