VideoSPatS: Video SPatiotemporal Splines for Disentangled Occlusion, Appearance and Motion Modeling and Editing

Juan Luis Gonzalez 1, Xu Yao 1, Alex Whelan 1, Kyle Olszewski 1, Hyeongwoo Kim 2, Pablo Garrido 1, 2 Imperial College London, [email protected] · 2025

Read VideoSPatS: Video SPatiotemporal Splines for Disentangled Occlusion, Appearance and Motion Modeling and Editing by Juan Luis Gonzalez 1, Xu Yao 1, Alex Whelan 1, Kyle Olszewski 1, Hyeongwoo Kim 2, Pablo Garrido 1, 2 Imperial College London, [email protected] (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-07146.txt.

Description

Juan Luis Gonzalez and five co-authors presented VideoSPatS in two thousand twenty-five, a way to model and edit video by splitting occlusion, look, and motion apart. Past methods map time and position to warping and colour, which tangles those three aspects. Their method maps positions into spline fields for space and colour, keeping motion and appearance separate.

Spline maths gives smooth flow over time and long-term consistency, both vital for believable edits. Extra branches peel a chosen occluder away from the main scene. Tests on standard sets plus the team's own talking-head clips show clean edits even with shadows, shine, and objects crossing the face.

Catalog record

CREATORJuan Luis Gonzalez 1, Xu Yao 1, Alex Whelan 1, Kyle Olszewski 1, Hyeongwoo Kim 2, Pablo Garrido 1, 2 Imperial College London, [email protected]
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-07146

Browse the full library · About the Tower of Babel