 There was a professor, his name is Makoto Nagao. He actually gave me a great advice and he said, you know, Mr. Kanade, research is not like that. Research is not like you pick up a topic and then just from sort of a shallow interest something and write a paper. It's not like that. You have to have some real goal and the topic must be the one which has some value so that when it works it can, you know, it has, it's useful. It has some real value in this, in the real society one way or another. Otherwise, it's not, it never be a good research. And then he told me that he showed me a big data set of human faces and then they collected 1,000, more than 1,000 facial image data, digital image. And at that time I always tell that researchers, researchers of computer vision had only one or two images in the lab. Only one or two. And you write a program and test your program with those one or two images and if they are successful, if the program is successful, then you write a paper. So you test only with one image. And I always tell that it's always successful because you can adjust the parameter so that it will succeed because you have only one. And then you write a paper. It's unbelievable. And if you test with your program with 10 images, we wrote very proudly in abstract with a very large-scale experiment with 10 images. It's a large-scale experiment. Today with 10 images, your paper will never be published. Large-scale means million today, probably. Or even more. At that time, 10 was large-scale. So he showed all those data, which was again stored in magnetic tape with 1,000 images. It's a 3 meter high pile of the reel of magnetic tape. And he showed me, here is the 1,000 facial images. And if you write a program which can successfully analyze these images in terms of locating eyes and so forth, then that alone will be a big accomplishment. So I started that and then because it was very concrete with the goal, I could very quickly do something real. And that changed my whole style of research at that time. In 1976, for the first time, I came to the United States and I visited Carnegie Mellon for one day to see if, as Dr. Newell suggested, to see if I feel it's an interesting place. And so I checked in on Friday and when I checked out Saturday morning and at the front, I said, oh, I'd like to check out. Tell me how much is it. And the guy said, free. You don't need to pay. I said, oh, why? And the guy said, well, Carnegie Mellon, your host, paid it. So I said, wow, this is a nice place to visit. And then I visited for one year. Yeah. However, about a couple of weeks into my visiting day, Rice already called me to his office. And he asked, Takeo, how much did I promise to pay you? And I said, 800. And he said, oh, that may be a little too little. He picked up the phone and he called a business manager. His name was Alan. Alan. Takeo salary should be 900. He hung up. And so from next month, salary became 900. A month later, he called me in to his office. Takeo, how much is your salary? I said, oh, you increased to 900. Oh, that's not good. He picked the phone. Alan. Takeo's salary should be 1000. He hung up, hands up. And he keeps doing it. And all the way, it became something like 2000 something. So I thought, well, this is not only a good place to visit, but this is a good place to stay. And you see, that's why I stayed all the way till 435 years after that. That's my favorite story. And always I said, well, when I look back, I remember it was always Saturday that he called me in. And whenever he picked up a phone and called Alan, he always, Alan was always at the end of the line. Can that be possible? Around the 2000 people of CBS visited me and I wasn't sure what they were interested. And they came me and they said, well, I saw the webpage of your virtualized reality. And they wanted, they said, well, what they want is to have a system that if there's a good, good play, then the same, the play can be viewed from all the 3360 direction. And that exactly one of the demonstration that we had from virtualized reality. So they said, well, in other words, they said, we want to do movie matrix like spinning replay. Now, the big difference in movie and football is that in the movie, you can tell the actor, please act here. And you can set that, set that up to focus at that particular point. However, in football, you don't know beforehand where a good play will occur. Therefore, you cannot have a static setup. So in order to do the similar thing, you have to move that focal point anywhere you want, depending on where is the play is occurring. Yeah. Therefore, what we did eventually was to replace all the cameras of a matrix setup by robot camera that can do pan and tilt and zoom so that all of them, according to the play, can see at the same time, same place. And that location is controlled to move by a cameraman, according to the play. And that is the iVision system. We had that idea and we told that idea to CBS. And then they said, okay, let's do it. And we had a contract. And indeed, we had only six months before from the time to start the project to the time of Super Bowl 2001, only six months. And it was a really highly pressured project, I recall. And in fact, we demonstrated a very primitive two or three camera system in New York in December. And we thought that we were confident that can be made bigger system. And we ordered 33 cameras from Japan from Mitsubishi Heavy Industry, each of which cost about $80,000. And the camera was a fairly high quality Sony camera with a big zoom lens, because you have to look, have a zoom view from stadium. And it was, we had a panic, because even one week before Super Bowl, the system that didn't work. And worse than that, actually, the circuit of robots began to explode. The condenser actually had somehow they explode. And robots one by one began not to work. And we were really panic. So we called my friend in Japan and then, oh, it's unbelievable. The engineer of Mitsubishi Heavy Industry actually flew to Florida, actually to look at the device. And together with him and our people found the reason and we quickly fixed it. And then barely made the system work a few to several days before the Super Bowl. And we, and then it was actually used in Super Bowl 35 2001 broadcast. And the game was not that good. So actually, it was so probably because the CBS didn't have other things to show. The system was used almost every play, spinning around and this and that.