Wednesday 8 April 2026 14:55
Wednesday, 8 April 2026, 14:55
PHOTO insait.ai
Font size
The Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) at Sofia University and streaming platform Netflix have developed a new artificial intelligence model capable of removing objects from video while realistically reconstructing the scene, the Bulgarian institute said.
The model, called VOID (Video Object and Interaction Deletion), goes beyond standard tools that simply fill in deleted areas. It analyses how objects interact within a scene and predicts what should happen once an element is removed. For instance, if someone holding an object is erased, the system can simulate how the object would naturally fall or move, adjusting shadows, motion, and other physical effects accordingly.
Unlike conventional video editing tools, which focus on visual consistency, VOID incorporates an understanding of physical causality — allowing it to recreate scenes as if they had originally been filmed without the removed object.
The technology is based on a large-scale video model and trained on synthetic datasets designed to simulate real-world interactions, enabling it to produce more realistic and coherent results across frames.
VOID could give film and video producers greater creative flexibility, reducing the need for costly reshoots and enabling more advanced post-production editing.
Editor: Miglena Ivanova
Edited and posted in English by E. Radkova
This publication was created by: Elizabeth Radkova