DeltaSplat: Iterative Gaussian Refinement for Pose-Free Feed-Forward 3D Gaussian Splatting
Pose-free feed-forward 3D Gaussian Splatting (3DGS) reconstructs a scene from sparse, unposed images in a single network pass, removing the need for camera calibration and per-scene optimization.
ProofPaper ↗
Key points
- To correct these errors, we introduce DeltaSplat, a lightweight Gaussian refinement module for pose-free feed-forward 3DGS.
- A 2D residual alone, however, underdetermines the 3D correction.
- DeltaSplat therefore conditions each update on per-pixel Plücker rays and rendered depth as a soft geometric prior.
- On DL3DV, DeltaSplat reaches 26.64 dB PSNR in the pose-free setting, improving its state-of-the-art backbone by 1.75 dB and surpassing even baselines supplied with ground-truth cameras; consistent gains hold across 6-24 views and all camera regimes.
Sources (1)
- [1]DeltaSplat: Iterative Gaussian Refinement for Pose-Free Feed-Forward 3D Gaussian SplattingarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 11:11 AM
Pose-free feed-forward 3D Gaussian Splatting (3DGS) reconstructs a scene from sparse, unposed images in a single network pass, removing the need for camera calibration and per-scene optimization.
To correct these errors, we introduce DeltaSplat, a lightweight Gaussian refinement module for pose-free feed-forward 3DGS.
Extractive summary: sentences quoted from the sources.