Lecture 17: Intro to Animation, Kinematics, Motion Capture (48)
yunerzxy
This article talks about motion capture in great details. My major take away from the lecture slide and the article is that motion capture could be a cost-saving solution to lots of scene simulations, but this data-driven approach also requires the correct techniques to capture, filter, analyze, assemble, and finally deliver the product that depicts real-world actions.
tyleryath
Recent work in pose estimation may soon become much better than traditional motion capture technology! This would get rid of the complexity associated with setting up and capturing a motion capture sequence. Here's a really cool dataset with some example footage of their technique in action: http://densepose.org/
sandykzhang
Similar to this, I read an article recently about college students from UW creating a set of sign language gloves that could translate sign language into spoken text. This seems like a similar application
emilyzhong
A really cool example of motion capture (with real time rendering, I believe) is Disney's "Turtle Talk with Crush" show, where the animated turtle's lips are animated to synch up with the voice actor's improvised responses (https://en.wikipedia.org/wiki/Turtle_Talk_with_Crush#Technology)
eliot1019
Also keep in mind that motion capture can also be used for facial animation and is commonly used in movies and video games like King Kong or Planet of the Apes.
AronisGod
Interesting fact, there are machine learning models which can train on motion captured data from professional dancers, then alter videos of other people standing still into videos of that person dancing professionally.
https://www.youtube.com/watch?v=PCBTZh41Ris
horrorsheep
Motion capture is good at face and fingers or subtle expressions.
michellebrier
Pretty cool article on this history of motion-capture movies: https://screencrush.com/motion-capture-movies/
This article talks about motion capture in great details. My major take away from the lecture slide and the article is that motion capture could be a cost-saving solution to lots of scene simulations, but this data-driven approach also requires the correct techniques to capture, filter, analyze, assemble, and finally deliver the product that depicts real-world actions.
Recent work in pose estimation may soon become much better than traditional motion capture technology! This would get rid of the complexity associated with setting up and capturing a motion capture sequence. Here's a really cool dataset with some example footage of their technique in action: http://densepose.org/
Similar to this, I read an article recently about college students from UW creating a set of sign language gloves that could translate sign language into spoken text. This seems like a similar application
A really cool example of motion capture (with real time rendering, I believe) is Disney's "Turtle Talk with Crush" show, where the animated turtle's lips are animated to synch up with the voice actor's improvised responses (https://en.wikipedia.org/wiki/Turtle_Talk_with_Crush#Technology)
Also keep in mind that motion capture can also be used for facial animation and is commonly used in movies and video games like King Kong or Planet of the Apes.
Interesting fact, there are machine learning models which can train on motion captured data from professional dancers, then alter videos of other people standing still into videos of that person dancing professionally. https://www.youtube.com/watch?v=PCBTZh41Ris
Motion capture is good at face and fingers or subtle expressions.
Pretty cool article on this history of motion-capture movies: https://screencrush.com/motion-capture-movies/