Netflix VOID Model Erases Objects From Video

Netflix has developed VOID (Video Object and Interaction Deletion), an AI model that removes objects from video footage and realistically inpaints how remaining objects would behave without them. It can transform a car crash scene into a clean road shot, erasing debris, smoke, and flames while generating physically plausible motion for remaining vehicles. The model is publicly available on Hugging Face and outperforms rivals like Runway in user testing, with VOID preferred 64.8 percent of the time versus Runway's 18.4 percent. Researchers say it excels at modeling complex physical dynamics following object removal.