A new open-source AI video editing tool developed by Netflix and Bulgarian researchers has demonstrated superior performance compared to commercial platforms like Runway, winning 64.8% of preference votes in direct human comparisons. The tool, called VOID (Video Object and Interaction Deletion), represents a fundamental shift in how AI approaches video editing by reconstructing scenes with physical consistency rather than simply erasing objects .

What Makes VOID Different From Existing Video Editing Tools?

Unlike conventional video editing software that removes objects and leaves empty space, VOID simulates how an entire scene should evolve after an object disappears. The system accounts for motion, object interactions, and environmental changes in ways that mimic real-world physics. This interaction-aware approach marks a departure from static editing toward dynamic scene understanding, an area where INSAIT researchers have been actively contributing .

The research team, led by researcher Saman Motamed, conducted a human preference study involving 25 participants to validate the tool's effectiveness. In direct comparisons, VOID was selected 64.8% of the time, significantly outperforming Runway, which received 18.4% preference in the same tests . This gap suggests that viewers notice and prefer the more physically realistic results that VOID produces.

How Does VOID Actually Work Under the Hood?

VOID integrates multiple AI components from leading technology organizations to achieve its results. The system combines several cutting-edge models and frameworks:

• Object Segmentation: Meta's SAM2 identifies and isolates objects that need to be removed from video frames
• Scene Analysis: Google's Gemini 3 Pro analyzes the broader context and relationships within each scene
• Video Generation: Alibaba's CogVideoX generates new video content to fill the space where objects were removed
• Quadmask Framework: A proprietary framework that allows the AI to understand what to remove and how surrounding elements should respond to that removal

At the core of VOID's innovation is the "quadmask" framework, which enables the AI to understand not just what to remove, but how the surrounding environment should respond. This is fundamentally different from tools that simply erase visual elements without considering the physics of the scene .

Why Is This a Big Deal for Video Creators and Studios?

The release of VOID on April 3 under an Apache 2.0 license marks Netflix Research's first publicly available AI tool, making advanced video editing capabilities accessible to creators beyond Netflix's internal teams. The open-source nature of the release means developers worldwide can build upon the technology, potentially accelerating innovation in AI-driven video editing .

For video production professionals, this development addresses a persistent pain point: the time-consuming and expensive process of removing unwanted objects from footage while maintaining visual consistency. Traditional methods require manual rotoscoping, green screen work, or expensive visual effects teams. VOID automates this process while producing results that human viewers prefer over existing commercial solutions .

The research was conducted using large-scale compute infrastructure and synthetic datasets, with INSAIT contributing to the development and experimentation process. INSAIT itself is a collaboration between Sofia University "St. Kliment Ohridski," ETH Zurich, and EPFL Lausanne, backed by $100 million in funding from the Bulgarian government and additional support from major tech companies including Amazon Web Services, Google, and DeepMind .

What Does This Mean for the Competitive Landscape?

The emergence of VOID as a superior alternative to commercial platforms like Runway signals a shift in the video AI market. While Runway has dominated discussions around AI video generation and editing, VOID's performance in human preference studies suggests that open-source, research-backed approaches can compete effectively with proprietary commercial tools. The fact that Netflix, one of the world's largest video production companies, chose to release this tool openly rather than keep it proprietary indicates confidence in the technology's robustness and potential impact .

This development also highlights the growing role of international research institutions in advancing AI capabilities. INSAIT's involvement demonstrates that breakthrough innovations in AI video editing are not limited to Silicon Valley or major tech company labs, but are emerging from research hubs across Europe and beyond. The Bulgarian-based institute's contribution to a Netflix-backed project underscores the globalization of AI research and development .

For creators evaluating video editing tools, VOID's public release offers a compelling alternative that combines academic rigor with practical performance. The tool's ability to reconstruct scenes with physical plausibility rather than relying on simple object removal suggests that the next generation of video editing software will prioritize scene understanding and realistic physics simulation over brute-force computational approaches.

" }