Netflix Releases VOID AI Model for Seamless Video Object Removal

• April 5, 2026

Netflix has introduced an advanced artificial intelligence model called VOID, designed to enhance video editing by seamlessly removing objects from footage and restoring the surrounding scene in a visually consistent manner. This technology addresses the challenge of erasing unwanted elements in video frames while ensuring that the remaining parts of the scene behave naturally without the removed objects.

AI-Powered Video Editing with VOID

Unlike traditional video editing tools that often leave noticeable artifacts or inconsistencies when objects are removed, Netflix’s VOID employs sophisticated AI techniques to not only eliminate specified objects but also to recreate the environment to reflect plausible interactions among the remaining elements. This includes reconstructing changes in lighting, shadows, and reflections that the deleted object had influenced.

The model was evaluated through a user study involving 25 participants, where it achieved a preference score of 64.8%, significantly outperforming its closest competitor, Runway, which received 18.4%. This comparison highlights VOID’s effectiveness in delivering more convincing video edits.

Netflix has made the VOID model accessible to the public via Hugging Face, enabling developers, filmmakers, and content creators to integrate this AI-powered video editing capability into their workflows. The release opens up new possibilities for creative editing, post-production workflows, and even user-generated content enhancement.

As video content continues to proliferate across streaming platforms and social media, tools like VOID serve as powerful enablers of high-quality production without the need for extensive manual editing or reshooting. This marks a step forward in applying AI to media technology, offering an efficient way to manipulate video scenes with minimal human intervention while maintaining authenticity and visual continuity.

Netflix’s new VOID AI model removes objects from videos and reconstructs scenes to maintain visual coherence, now available on Hugging Face.