This invention describes a system for adding a virtual character or item into a video of a real person. It works by analyzing the person's movement in the video, specifically identifying a body joint that moves the least, and using that as a stable anchor point for placing and moving the virtual item. The system can also adjust the virtual item's size to match the person's body parts.
Why it matters: Since 2023, advancements in real-time AI-driven pose estimation and augmented reality rendering on consumer hardware have made the specific tracking and display methods claimed more accessible and performant. This could simplify the implementation of robust skeletal joint identification and virtual item integration.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI