# Real-Time Facial Expression Transfer | Two Minute Papers #21 ## Метаданные - **Канал:** Two Minute Papers - **YouTube:** https://www.youtube.com/watch?v=mkI6qfpEJmI - **Дата:** 30.10.2015 - **Длительность:** 2:04 - **Просмотры:** 13,375 ## Описание In computer animation, animating human faces is an art itself, but transferring expressions from one human to someone else is an even more complex task. One has to take into consideration the geometry, the reflectance properties, pose, and the illumination of both faces, and make sure that mouth movements and wrinkles are transferred properly. The fact that the human eye is very keen on catching artificial changes makes the problem even more difficult. This paper describes a real-time solution to this animation problem. ______________________________________ The paper "Real-time Expression Transfer for Facial Reenactment" is available here: http://graphics.stanford.edu/~niessner/thies2015realtime.html Recommended for you: ALL Two Minute Papers episodes! :) - https://www.youtube.com/playlist?list=PLujxSBD-JXgnqDD1n-V30pKtp6Q886x7e Subscribe if you would like to see more of these! - http://www.youtube.com/subscription_center?add_user=keeroyz The thumbnail image was created by tommerton2010 (CC BY 2.0) - Changes have been made to it (eye color, flip, background). https://flic.kr/p/9d8ApH Splash screen/thumbnail design: Felícia Fehér - http://felicia.hu Károly Zsolnai-Fehér's links: Patreon → https://www.patreon.com/TwoMinutePapers Facebook → https://www.facebook.com/TwoMinutePapers/ Twitter → https://twitter.com/karoly_zsolnai Web → https://cg.tuwien.ac.at/~zsolnai/ ## Содержание ### [0:00](https://www.youtube.com/watch?v=mkI6qfpEJmI) Dear fellow scholars, this is two minute papers with Kohaa. Today we are going to talk about a great algorithm that takes the facial expression of one human and transfers it onto someone else. First ### [0:12](https://www.youtube.com/watch?v=mkI6qfpEJmI&t=12s) Tracking Pipeline there is a calibration step where the algorithm tries to capture the geometry and the reflectance properties of both faces. The expression transfer comes after this which is fraught with difficulties. It has to be able to deal with changes in the geometry, the reflectance properties of the face, the illumination in the room, and finally changes in pose and expressions. All of this at the same time and with a ### [0:35](https://www.youtube.com/watch?v=mkI6qfpEJmI&t=35s) Real-time Facial Reenactment negligible time delay. The difficulty of the problem is further magnified by the fact that we humans know really well how a human face is meant to move. Therefore, even the slightest inaccuracies are very easily caught by our eyes. Add this to the fact that one has to move details like additional wrinkles to a foreign face correctly and it's easy to see that this is an incredibly challenging problem. And the resulting technique not only does the expression transfer quite well, but is also robust for lighting changes. However, it is not robust for ### [1:10](https://www.youtube.com/watch?v=mkI6qfpEJmI&t=70s) Tracking Failures occlusions, meaning that errors should be expected when something gets in the way. Problems also arise if the face is turned away from the camera, but the algorithm recovers from these erroneous states rapidly. What's even better, if you use this technique, you can also cut back on your plastic surgery and hair plantation costs. How cool is that? This new technique promises tons of new possibilities beyond the obvious impersonation and reenactment fund for the motion picture industry. The authors proposed the following in the paper. Imagine another setting in which you could reenact a professionally captured video of somebody in business attire with a new real-time face capture of yourself sitting in casual clothing on your sofa. Hell yeah. Thanks for watching and I'll see you next time. --- *Источник: https://ekstraktznaniy.ru/video/14930*