AdaptiveSplat:Texture Aware Controllable 3D Gaussian Allocation for Feed-Forward Reconstruction
About
Current feed-forward 3D reconstruction methods predict pixel aligned Gaussian primitives, resulting in highly redundant representations. A natural solution is to prune the redundant Gaussians, but naive pruning introduces severe artifacts and often requires inference time fine-tuning, breaking the feed-forward paradigm. Based on previous works, high frequency regions require more Gaussian primitives, while low frequency regions can be represented with significantly fewer primitives. Motivated by this, we propose a novel approach to explicitly control the number of Gaussians by leveraging local texture information. Our approach achieves this through three key components: (1) texture estimation to capture spatial variation in scene detail, (2) texture-aware pruning that removes redundant Gaussians from low frequency regions, and (3) an adaptive Gaussian head that predicts the modified attributes of the retained primitives without breaking the feed-forward paradigm. Experiments on RE10K, ACID, DL3DV, Tanks and Temples, and DTU demonstrate the effectiveness of our approach, while ablation studies validate the contributions of its key components.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Novel View Synthesis | RE10K | SSIM74 | 345 | |
| Novel View Synthesis | ACID | PSNR22.55 | 175 | |
| Novel View Synthesis | DTU | PSNR15.23 | 154 | |
| Novel View Synthesis | DTU (test) | PSNR16.19 | 119 | |
| Novel View Synthesis | ACID (test) | PSNR22.55 | 113 | |
| Novel View Synthesis | DL3DV 6view | PSNR20.448 | 46 | |
| Novel View Synthesis | DL3DV 32 views | PSNR17.677 | 37 | |
| View Synthesis | Re10K (test) | PSNR20.74 | 30 | |
| Novel View Synthesis | DL3DV 9 views | PSNR19.678 | 24 | |
| Novel View Synthesis | DL3DV 16 Views | PSNR19.125 | 24 |