
New Method Fixes Temporal Instability in Single-Camera 3D Video Reconstruction
Lift4D, a video harmonization technique, corrects the frame-to-frame geometric inconsistencies—scale drift, flickering surfaces, temporal discontinuities—that emerge when single-view depth estimators are applied sequentially to monocular video. By reconciling per-frame outputs into a coherent 4D representation, Lift4D enables reliable scene reconstruction from ordinary smartphone or consumer footage without multi-camera rigs or studio setups. The preprint remains unvalidated against diverse real-world footage.
Published