Search papers, labs, and topics across Lattice.
ATSplat introduces a novel feed-forward 3D Gaussian Splatting framework that enhances adaptive capacity allocation by utilizing Adaptive 3D Tokens, allowing for a more efficient representation of complex scenes. Unlike traditional methods that rely on pixel-aligned formulations, ATSplat decouples the placement of primitives from input image grids, enabling a sparse-to-adaptive approach that focuses on high-uncertainty regions. Experimental results demonstrate that ATSplat achieves state-of-the-art rendering quality while significantly reducing the number of Gaussians, achieving over 5.7 times fewer Gaussians than dense methods, and rendering high-quality novel views at impressive speeds.
ATSplat achieves over 5.7 times reduction in Gaussian primitives while delivering state-of-the-art rendering quality and real-time performance.
3D Gaussian Splatting (3DGS) achieves high-quality novel-view synthesis by optimizing freely placed primitives in 3D and adaptively densifying them in under-reconstructed regions. However, this scene-adaptive capacity allocation is largely lost in existing feed-forward 3DGS methods, which commonly regress Gaussians at input pixels and lift them along camera rays. Such pixel-aligned formulations make the number and placement of primitives depend on image resolution and input viewpoints rather than scene complexity, resulting in dense and often redundant Gaussian sets. We present ATSplat, a feed-forward 3DGS framework that restores the adaptive allocation capability of 3DGS optimization through Adaptive 3D Tokens. ATSplat first lifts coarse patch-level depth and camera cues into sparse 3D anchor tokens, forming a compact scaffold of the scene. Each token is then regressed into local Gaussians with learnable 3D offsets, decoupling primitive placement from input image grids. An Adaptive Token Expansion module predicts a token-level uncertainty score, supervised by rendering error maps, and selectively expands high-uncertainty tokens through learnable expansion layers. This sparse-to-adaptive formulation enables ATSplat to concentrate primitives in challenging regions while maintaining a compact representation. Experiments on two representative datasets, RealEstate10K and DL3DV, show that ATSplat achieves state-of-the-art rendering quality while reducing the number of Gaussians by more than $5.7\times$ compared with dense feed-forward 3DGS methods. From 12 input images at $512 \times 960$ resolution, ATSplat completes reconstruction in less than a second using a single commercial GPU, and renders high-quality novel views at 1136 FPS ($512 \times 960$) with only 311K Gaussians.
NUS