Creat3r: Confidence Reaggregation for Exploration-aware Active 3D Reconstruction

We present Creat3r, an iterative next-best-view (NBV) selection framework for efficient, high-quality 3D reconstruction. Starting from a small seed set of image-pose pairs, Creat3r repeatedly selects the most informative next camera pose. After each pose is chosen, the corresponding image is acquired and added to the multi-view set to update a 3DGS reconstruction. To guide selection, Creat3r constructs an intermediate point cloud and estimates reconstruction reliability via a novel 3D confidence field, which is projected to candidate poses through Gaussian projection to produce 2D confidence and exploration maps. These maps balance exploitation of reliable regions and exploration of uncertain or unseen areas under computational constraints. Experiments with standard 3DGS show that Creat3r consistently outperforms baselines in novel view synthesis and surface reconstruction, achieving higher SSIM and F1 scores with fewer views.

When building a 3D digital model of a real-world scene, an agent needs to take photos from multiple angles. However, taking too many random pictures is inefficient, while taking too few leaves gaps in the model. We introduce Creat3r, an AI algorithm that acts like a smart photographer. Creat3r continuously estimates what the 3D model looks like so far and identifies which areas are still blurry or uncertain. It then decides exactly where to move next to get the most valuable new information. Creat3r works seamlessly with different 3D modeling technologies and creates highly accurate, complete 3D digital twins with fewer images.