0
2022.
Neural RGB-D Surface Reconstruction – Supplemental Document –
Figure 1. Our method obtains a high-quality 3D reconstruction from an RGB-D input sequence by training a multi-layer perceptron. In comparison to state-of-the-art methods like BundleFusion [3] or the theoretical NeRF [9] with additional depth constraints, our approach results in cleaner and more complete reconstructions. As can be seen, the pose optimization of our approach is key to resolving misalignment artifacts.