
In his video titled “Meta’s SAM 3: AI Vision just got a HUGE UPGRADE (FREE),” published on December 10, 2025, Matthew Berman explores Meta’s latest AI model, SAM 3. This open-source model revolutionizes video editing by automating a process known as rotoscoping, which traditionally involves manually segmenting elements in a video. With SAM 3, users can now perform segmentation tasks using simple text prompts—highlighting objects merely by describing them. Berman demonstrates this capability by showing how the model can highlight dogs, zebras, and other objects in videos, a testament to its accuracy and ease of use. The feature also extends to segmenting animated objects like skateboards or taxis with just a click. He praises the model’s intelligence in distinguishing between similar objects, such as different flavors of ice cream, showcasing its advanced understanding of visual elements.
However, while Berman is enthusiastic about SAM 3’s capabilities, there are some areas where the technology might face challenges. The promise of completely automating rotoscoping in complex scenes may not account for nuanced scenarios where manual intervention is needed. An AI’s inability to perceive context as humans do might result in occasional misclassifications or oversights.
Yet, Berman’s excitement is palpable as he guides viewers through SAM 3’s user-friendly interface, highlighting features like object previews, customizable effects, and downloadable templates for common tasks like pixelating license plates. He notes its utility not just for video editors but also for animators and game developers looking to streamline workflows.
SAM 3’s potential extends beyond creative fields. Berman envisions its use in robotics, where the model’s segmentation ability could enable machines to navigate more intelligently and safely. He concludes with an invitation to viewers to experiment with SAM 3 themselves, emphasizing its open-source nature as a core strength that democratizes access to advanced video editing tools. Despite the potential hurdles, this new chapter in AI-driven video editing is set to empower creators and technologists alike to innovate and expand their horizons.