Hardware
ZipMap: Linear-Time Stateful 3D Reconstruction via Test-Time Training
arXiv:2603.04385v3 Announce Type: replace Abstract: Feed-forward transformer models have driven rapid progress in 3D vision, but state-of-the-art methods such as VGGT and pi^3 have a computational
arXiv:2603.04385v3 Announce Type: replace Abstract: Feed-forward transformer models have driven rapid progress in 3D vision, but state-of-the-art methods such as VGGT and pi^3 have a computational cost that scales quadratically with the number of input images, making them inefficient when applied to large image collections. Sequential-reconstruction approaches reduce this cost but sacrifice reconstruction quality. We introduce ZipMap, a stateful feed-forward model that achieves linear-time, bidirectional 3D reconstruction while matching or surpassing the accuracy of quadratic-time methods. ZipMap employs test-time training layers to zip an entire image collection into a compact hidden scene state in a single forward pass, enabling reconstruction of over 700 frames in under 10 seconds on a single H100 GPU, more than 20imes faster than state-of-the-art methods such as VGGT. Moreover, we demonstrate the benefits of having a stateful representation in real-time scene-state querying and its extension to sequential streaming reconstruction.
Related
- Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction
- AnchorSplat: Feed-Forward 3D Gaussian Splatting with 3D Geometric Priors
- Self-Improving 4D Perception via Self-Distillation
- MV-SAM3D: Adaptive Multi-View Fusion for Layout-Aware 3D Generation
- Splats under Pressure: Exploring Performance-Energy Trade-offs in Real-Time 3D Gaussian Splatting under Constrained GPU Budgets
Source: arXiv cs.CV | 2026-04-10