Using Google Cloud, Gemini and Deepmind research, developers have developed an AI tool that extracts 3D biomechanical data from standard video. The AI was used to measure the rotational degree of tricks completed by snowboarders and skiers. It essentially turns any camera into a motion-capture system without the use of any sensors on the athlete.

Traditional trick names like a 1440 use 180 degree increments as shorthand for trick difficulty, the AI measures the true geometric rotation of the body. For instance, athlete Shawn White’s 1440 was measured by AI as actually 1122 degree of physical rotation. The data reveals a gap between the trick’s name and the actual degrees rotated. A smaller number of degrees usually indicates trick mastery as the athlete has found a more efficient axis to complete the trick, allowing for more style and a cleaner landing.

This new metric will be used in coaching and broadcast graphics, providing a scientific way to compare different athletes’ techniques and helps determine the physical limits of what is possible as the sport pushes towards even more complex tricks.