VGG-T$^3$: Offline Feed-Forward 3D Reconstruction at Scale

Forward time (s) OOM VGGT FastVGGT SparseVGGT Ours (b) Num. images vs. inference time.

Figure 1. Reconstructing Rome landmarks with 1-minute time budget.

We present VGG-T 3 , an offline feed-forward 3D reconstruction method that scales linearly w.r.t. input views (Fig. 1b). As a result, we can reconstruct large scenes from a large number of unposed input views, such as landmarks from tourist-sourced images, in less than a minute via single forward pass (Fig. 1a).