Difference between revisions of "First-Hand:My Recollections of the Development of the Glowing Hockey Puck"
m (First-Hand:Ice Hockey & Broadcast Technology: The Origin of the Fox Sports Innovation moved to First-Hand:My Recollections of the Development of the Glowing Hockey Puck)
Revision as of 15:24, 4 March 2010
Contributed by Stan Honey
Origin of the Idea
NewsCorp was then a $10 billion worldwide media company with newspapers and book publishers and TV broadcasters etc when it acquired ETAK. I thus became the CEO of a tiny operating division within NewsCorp. At corporate management conferences, I soon got to know the other chief executives. As it turned out, I was the only engineer that many of them knew. Soon the CEO’s running other operating divisions started to call me and ask questions about technology, often about encryption, digitization, compression, the internet, conditional access, copy-protection etc. Some of them had projects in which they were signing technology contracts to modernize their business. Microsoft was calling on them to propose content licensing and electronic distribution deals and they wanted to be sure that they didn’t get blinded by the technology. Fox wanted to talk about how technology could be used to make television more interesting or cheaper to produce. So I began spending half of my time answering technology questions and doing in house consulting, within News Corporation, for these other much bigger operating divisions. I think that is why Rupert formalized that role and asked me to become Exec VP Technology for NewsCorp.
At that time, David Hill was running Fox Sports. David asked me to meet with him every week or so to brief him on technologies that could potentially have an impact on the televising of sports. In one of those meetings I briefed David on the fact that it would be possible to insert fake billboards in televised sporting events. He responded -- and this is a refrain I was to hear many, many times over the years that I worked with David – “this is possibly the stupidest thing I’ve ever heard.” But then he would ask, “but if you could do that, could you do this (other thing) that would be useful? In this particular instance David asked whether if we could insert things in real video that weren’t there, could we do something useful like highlight the position of a hockey puck? I told him that I’d think about it. I then called him back a week later and replied “yeah, we could do it, but you probably can’t afford it.” I mentioned that we had tracked tougher things at SRI and the group of guys that I worked with at SRI and then subsequently at ETAK could build a system to track and highlight a puck in live broadcast TV. When David asked what it would cost and I estimated $2 million and two years. I had spent a week figuring out the estimate and thinking about a couple of possible ways to do it in order to frame the cost.
David responded, “send me a memo.” So I sent him an email saying that it would cost $2 million and take two years, and then just an hour or two after I sent that email, I got a call from Rupert, and Rupert said “David tells me you can track and highlight a hockey puck for $2 million in two years.” I said “that’s right” and Rupert said “do it.” Rupert added, “if anybody asks where the money’s going, have them call me”. Those are the important words at NewsCorp. I had $2 million and wondered what I had gotten myself into. Fortunately I was able to drag Rick Cavallaro and a team of other key engineers into the project.
Once you’re an experienced hockey fan, watching the game is all about the gestalt of where all the players are and the flow of the game. It’s not really that important that you see the puck all the time because you see the big picture and you kind of know where the puck is anyway. But for somebody who’s not a hockey fan, the game doesn’t make any sense if they can’t see the puck. There’s just all these guys skating around, and it’s hard for an inexperienced, potential hockey fan to figure it out. It then becomes hard for them to become a hockey fan without having the initial ability to see the puck.
David figured that the die-hard hockey fans are going to watch no matter what you do, but how do you get the non-hockey fans to watch when they can’t see the puck and they don’t get it? So David concluded that if we could track and highlight the puck, Fox would be able to increase viewership of Hockey. David correctly predicted that it would annoy the diehard fans, but they were going to watch anyway. He had the brilliance to figure out not to just do billboard ads but to do something that would help folks understand a sport.
Setting up the Team
The first thing I did was to grab the same group of guys with whom I’d been working with at SRI and then later at ETAK. We got the core group together: Rick Cavallaro was the central and most important guy; along with Terry O’Brien. I told David that I could pull together a team that would do the technical part of tracking of the puck and generating a computer graphic 60 times a second highlighting the puck in every field of video. But I also told him that I did not know anything about field broadcast TV operations. We didn’t know how to integrate with the field broadcast operation. David solved that problem by introducing me to Jerry Gepner, who was then head of Operations for Fox Sports. Jerry played a key role in the Puck system. Later he, Bill Squadron, and I became the co-founders of Sportvision.
Jerry is a guy who came up through the hawse-hole in TV operations, working on production trucks, but Jerry is very interested in technology. Unlike a lot of TV field engineers, Jerry was an advocate for new stuff. So he was just delighted to be involved and was enormously helpful. He presence in the project enabled us to build a system that looked like a TV system, smelled like a TV system, felt like a TV system, and worked like a TV system. So when we did show up at the All Star game in ’96, our system fit right in with the broadcast operations and the other TV field operations engineers figured that we were TV guys because due to Jerry’s input, our system was clearly designed properly to integrate into TV field operations.
The puck project was definitely going to be challenging. It had never been attempted before as far as we knew. It was like being back at SRI, where you never were asked to solve a problem that you knew how to solve or that anybody else has solved before. At SRI we were asked to tackle problems that no one had solved. My whole career has been spent working on things that nobody had done before. That’s part of the excitement. There is the adrenaline rush of the challenge and the fear that you’ll fail, and the elation when you do get it to work. The key to succeeding, and to enjoying the challenge, is working with a highly skilled team.
Choosing the Technical Approach
The Radar Option
One alternative that we investigated was to put a FMCW radar transponder in the puck. That plan was to have multiple FMCW radars in the arena and a frequency translating transponder in the puck. It was similar to some tracking systems that we had worked on at SRI. The problem was the high cost of the pucks. 30 pucks can get used in a game and they are not returned. We could not afford to lose so many of expensive pucks in a game. Although we got a prototype of the radar approach to work, we abandoned it. We had also considered retro reflectors on the puck, illuminating them with infrared and detecting the reflection. That didn’t work very well and was quickly dropped.
Infrared Transmitter in the Puck
The approach that we ultimately used was having a puck that emitted 100-microsecond pulses of infrared 30 times a second. We placed a number of optical sensors around the rink to measure the angle to the puck. We used a number of tricks to get the signal to clutter ratio quite high. We shuttered the IR sensors so that they only had their (electronic) shutters open during the 100 microsecond pulse that was coming from the puck. We had cavity interference filters over the sensor lenses so they were only sensitive to the particular very narrow wavelength of the infrared emission from the puck. We ended up with very high signal-to-clutter and were able to very clearly measure the angle to the puck from each of the sensors. It was six months into the project before we stopped work on the other approaches and committed to the infrared emitter approach. The work we had done on some of those other approaches was helpful to us because we gained insight that we were subsequently able to use in the development of the final system. For example, some of the experiments we did on the infrared retro reflecting gave us better insight into infrared in an arena.
Integrating the TV cameras
We knew that the system had to know the field of view for every field of video from any camera that was going to include the highlight and so we knew that the system had to measure the pan, tilt, zoom, and focus of every augmented camera very accurately, 30 times a second. Jerry informed us that all field cameramen want to use the Vinton Vector 70 head. So then it was a matter of buying a Vinton Vector 70 head and figuring out how to add the sensors to it. Terry Obrien sorted that out. He is the mechanical engineer who also designed the mechanical parts of the Etak Navigator, as well as the puck mechanics and circuit board
During the 11 months of engineering development, we were focused on getting the puck tracking system to work in time for the Jan 1996 All Star game. We weren’t necessarily thinking about having it run inexpensively. So in the original hockey system, there was an industrial PC in an industrial enclosure next to each camera with multiple wires going to the camera and to the lens and to the pan-tilt head making all of the measurements, and several twisted pairs that went from each camera site down to the puck truck, as we called it. Later, as the system got better engineered for the Yellow First Down Line, there wasn’t any computer up near the camera. There was just a little box of sensor and telemetry electronics that was bolted in the front of the head and we didn’t have to run any extra wires. The telemetry all took place through the camera triax and the data was passed via an unused mic channel. As the years went on later at Sportvision, we reengineered the system to be much smaller, cheaper, easier, but the objective of the original Puck system was to get it to work.
Demos for Fox
We did a number of demonstrations during the development for David Hill and for Rupert Murdoch, but our first demo for David, Rupert, and the NHL was more flash than substance. Pun intended unfortunately.
The first demonstration for the NHL took place in an interior conference room that had no external windows, and had no incandescent lights. I.e. there were no sources of infrared in the room. We built a puck that had an accelerometer, battery, and IR emitters that turned on for a minute after every time the Puck received a shock. Unlike future pucks, this puck emitted CW IR for a minute, not 100 microsecond pulses. We set up a very sensitive camera in the room that was sensitive only to IR, and ran the output to the TV monitor in the room. We turned up the sensitivity of the camera such that the view of the room looked very normal in the monitor. Then when somebody took our sample puck and slammed it on the table, the puck looked the same to the eye, but if you looked in the monitor the puck was so bright that it saturated the camera and lit up everything near it. It was an amazing view. In the monitor it looked like the guy holding the puck would burn his hand off, but if you looked right at the puck you couldn’t see anything abnormal.
Obviously this approach wouldn’t work for the actual system, but it was an opportunity to talk about putting IR emitters in the puck with the NHL and give them some feel for the technical approach.
As we got into actual development, we worked in San Jose at the Sharks Arena. The San Jose Sharks arena was the most geographically convenient hockey arena for us. We spent weeks in the rafters there.. The Sharks were hugely supportive of the whole project. There were probably four or five times David Hill and his team from Fox Sports would come by San Jose to see a demo. We would show them what we had at the time. Several times we hired a few hockey players to go swat pucks around so David could get a feel for the system.
And then there was the time when Rick and I had to do some unorthodox testing of the puck, and settle a bets. Rick, who was an aeronautical engineer by training, and I, were interested in the terminal velocity of a hockey puck. I can no longer recall the reasons for this interest. I do remember that we were trying to settle a bet probably for $1 but we take these $1 bets very seriously and cherish those dollars. I also recall the odd experimental methodology that we adopted. We figured that to measure the terminal velocity of a falling puck, we would dangle a puck from a string suspended from a stick that we stuck outside a car window. Then we’d drive fast enough for the puck to dangle at 45 degrees from the vertical. At that car speed, the force of gravity on the puck was the same as the force of the air, and so the terminal velocity of a falling puck would be the speed of the car. Rick and I agreed on the methodology of the test, strung up a puck, string, and stick, and headed of to tear over the Dumbarton Bridge during the uncrowded time of the mid-morning. The terminal velocity turned out to be 75 mph or so. It got scary, however, when the puck came off of the string and went bounding along the highway. Fortunately it bounced over the side of the bridge into San Francisco Bay before hitting another car. Unfortunately, that was not the stupidest thing that Rick and I have done.
David and his team had a bunch of decisions to make, like what did they want the system to do? What did they want it to look like? There were a number of different graphic treatments that we implemented. It ultimately ended up being a red rocket trail and a blue fuzz ball. That wasn’t our part of the system to design. David is the storyteller, whereas the technical team, we just make it work. So for the most part, the ideas of what to do and how to make it look come out of conversations with storytellers like David. Although later, Terry Obrien suggested some features like the X-Ray view that allowed viewers to see where the puck was when it was below the boards. That feature made it to air.
There were some very productive exchanges when we got my team working with guys like David. I’d say” hey, just for you to think about, you should be aware you can do this kind of stuff”, and then I’d sketch out an idea. And David would always say “that’s the stupidest thing I’ve ever heard, but if you can do that, could you do this other thing” and we would have an incredibly fruitful conversation. After an hour or so of kicking things around David would often end up stating, “well now that’s something I could really use.” and then off we’d go. Many of the systems that we ultimately developed at Sportvision came from brainstorming sessions like these with David Hill at Fox.
Week of the All-Star Game
A week before the All-Star game we had problems with incandescent lights in the Boston arena. We had to solve these problems by changing our software to mask out areas of specular reflection of the incandescent lights in our IR sensors. Just before the start of the All-Star game we had intermittent problems with our telemetry from our sensors in the arena. At the last minute we found that the interference was coming from the light dimmers in our truck causing interference to our telemetry. Fortunately we found the problem before the game started, and it worked fine.
The puck tracking system ended up costing $2.2 million So we ended up 10% over budget, but it worked the night of the All Star game, which is the original scheduled night of introduction. We hit the time spot-on. It was hairy. It did work perfectly that first All Star game, but we had some very scary problems up to a few hours before the game.
Fox’s PR Effort
The thing that added pressure to the Hockey system, and later to Sportvision, was the high public visibility. The publicity buildup to the All-Star game also added a lot of pressure on us, but that was the deal. We said we could do it in on time and for $2 million. They took that at face value and they assumed it was going to work at the All Star game.
That is the assumption in the TV production business. If you say that something will work for the All Star Game, or the Super Bowl, the customer doesn’t bother with clauses in the contract dealing with delays. It has to work on time.
The publicity surrounding the introduction of the puck tracking system was fascinating. The die-hard hockey fans hated it. But they talked about it a lot. There was a ton of controversy about it. We started collecting clippings, and I have a set of them and I’m sure Rick has a set. The clippings were an inch and a half thick. David’s conclusion was that you couldn’t pay for that kind of visibility and promotion. There was even a David Letterman show where Letterman did a spoof of the thing. It was fabulous, and created tons of publicity for Fox.
David’s point was that huge publicity was very valuable. You can take advantage of the controversy. That’s worth millions. David believed that the press was worth the $2 million right there to get that much visibility. David figured you could risk annoying the die-hard fans; you’re going to keep them anyway. The key was attracting new fans. The rating for hockey those three years that they used the puck on the A game every week was the highest rating hockey ever had and has ever had since, with the exception of that Olympic game against the Soviet Union. Part of it was, I think, that the new fans could get more involved, and part of it was all the controversy.
The system was used in televised hockey until Fox lost the broadcast rights, I believe 3 years later. Someday it will return, possibly to the dismay again of the die-hard fans.