This invention describes a way to make a person in one video move exactly like a different person in another video, all happening live. It works by taking the visual look of the first person and applying the movements of the second person to it, creating a new video where the first person performs the second person's actions. The system extracts detailed visual information and motion data using machine learning techniques.
Why it matters: Since 2021, generative AI and real-time computer vision models have advanced significantly, making the "real-time" aspect and the quality of motion transfer more achievable and less computationally prohibitive.
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