>>[music playing] This program is brought to you by Stanford University. Please visit us at Stanford.edu. Siddharth Batra: Our technology empowers anyone with a video camera to very easily, automatically, place pictures, videos, and advertisements within their home videos and then they can upload them onto the web. Anybody with a home video, like the one in front of you, can, within a few seconds and a couple of clicks of the mouse, select the surface they want to replace with their picture advertisement, and upload this video to our web site. After which, within a few minutes of processing, they can see– they can see a variety of ads on that content. Users can take their homemade videos, place ads in them, and put them online. So, in a way similar to what Google AdSense did with bloggers, people can now sit at home, try and make videos, and get money for that.>>Ampere application allows people to create novel visual effects in their videos. For example, if said person has a video of their house and they want to see how would a painting look on their wall, they could actually go ahead and use ampere technology to put that new painting on the wall and decide and get a feeling for how it would look. Siddharth Batra: The technologies in this form exist in professional movie houses or, umm, movie studios. But with the use of our technology, it allows amateurs and normal people on the street with a camera and home videos to very easily edit– edit their home videos, place ads on them, take pictures and…>>[cheering] You get a single image and our program can figure out which part belongs in front and which parts are in the back– in the background. We have extended this technology to videos and the accelerometer has to figure out which parts of the scene are actually in the foreground, that is the first sense so that the ad could be placed on the wall behind the person. Siddharth Batra: This technology is called the 3-D surface tracker technology. Umm, it starts off that we’re given a video and a surface in the video that we would like to replace with some other content. You begin by, umm, analyzing the video for all sorts of camera motions. Like, if the camera panning, or zooming in, or is the camera itself moving. So, for each pixel in that surface, we build a model of how it looks like and what our belief is what it will look like in the next frame. If said lighting begins to change with the motion of the video, or the sun, or the shadows, then we keep a belief of what it will look like in the next frame and this is how we track with very high subpixel accuracy. You can have multiple copies of the same video and if, suppose, you wanna show it to your parents, you can put a nice poster there of Albert Einstein. But if you wanna show it to your friends you can have a Playboy poster there. Anything you want. Yeah.>>[laughing]>>[music playing] The proceeding program is copyrighted by Stanford University. Please visit us at Stanford.edu.