ESPN Brews 3D Melting Pot With Variety of Vendors at ESPN 3D Tech Days
With upwards of a dozen vendors present and a host of 3D technologies undergoing live tests for the first time, ESPN’s 3D Tech Days last week at Rentschler Field in Hartford, CT, was a monumental event in the evolution of live 3D television. However, to gauge just how groundbreaking the series of 3D tests were, you have to go to one of the industry leaders. When asked if it were fair to say that the ESPN event was the biggest 3D test that had ever been done for television, 3D guru and PACE CEO Vince Pace replied, “Absolutely. Just look at all the interaction here. You have so many people coming from different venues of [3D production] and talking. This has got to be the biggest. It must be, because I’ve been to pretty much all of them.”
In an effort to further develop best practices for its 3D coverage of college football next fall, ESPN brought in the Connecticut Thunder, a semi-pro football team from the New England Football League, to run plays for about three hours on May 27 and 28 at Rentschler while the network experimented with its 3D coverage. For the event, ESPN brought in several vendors to test out 3D products as well as most of its 3D A-team in the truck, including producer Josh Hoffman and director Doug Holmes. The vendors invited were NEP, Fletcher, Broadcast Sports Inc. (BSI), PACE, SkyCam, SportVision, PVI Virtual Media Services, Robo-Vision, Woods Hole Oceanographic Institution, and MobileSat Communications.
According to ESPN VP of Event Operations Chris Calcinari, NEP’s SuperShooter 3D truck is currently being reboxed, while NEP’s second 3D truck is under construction and will be ready for ESPN’s 3D Home Run Derby show in July. As a result, ESPN used SuperShooter 9, typically a 2D truck, with a few minor 3D adjustments. The “production” featured six 3D camera rigs, most of which set up in non-traditional positions.
Among the firsts being tested at 3D Tech Days: a variety of experimental camera angles, a new camera setup loosely referred to as a MastCam, a 3D super-slo-mo camera system from Fletcher, wireless RF 3D camera systems, a 2D/3D Skycam, a remotely operated 3D pan-bar camera system, a first-and-10 line for 3D, and discrete transmission for both left eye and right eye back to ESPN’s headquarters in Bristol, CT.
Old Cameras, New Angles
The MastCam features a PACE 3D rig atop a tall, vertical pole similar to a sailing mast. The pole is attached to a motor-powered cart, which allows it to move up and down the sidelines. The robotic camera is controlled remotely by a camera operator using an off-site pan-bar system. ESPN sees this system as a potential game-camera angle that would help to avoid the necessary seat kills caused by setting up 3D rigs in the stands.
“This allows us to get closer to the action without killing seats,” says Calcinari. “One of the thoughts is that this cart would have the ability to travel up and down the sidelines and we would actually do a play-by-play camera off of this, using the Robo-Vision or Fletcher pan-bar system to control the heads. That way, this camera wouldn’t have to be in the stands.”
The goal posts have presented a serious issue for 3D productions. The traditional behind-the-end-zone position is unusable for 3D productions, because the posts leap into the foreground, creating an extremely uncomfortable 3D experience for the viewer. As an alternative, ESPN experimented with a PACE camera rig directly below the crossbar of the end zone. The rig at Rentschler was far too large to be used for an actual live game, but ESPN says that, if it works, they can switch it out for an ultra-small 3D rig and get it approved for game coverage.
“We can’t use the normal behind-the-end-zone position because we have to look through the bars, we know that,” says Stephen J. Raymond, coordinating technical manager, ESPN. “And our [TV partners] won’t let us hang anything on [the front of the cross bar]. But we can hang anything [below the crossbar] or on the [base pole] as long as it’s not anything too big. We definitely feel like we can get approval for this.”
Fletcher rolled out the 3D super-slo-mo camera rig it demonstrated at NAB. The rig, featuring two cameras shooting at 1,000 frames per second (fps), was positioned on the field level at the back of one end zone. Although tests had produced excellent results, according to Fletcher VP Dan Grainge, the rig had yet to be tested in a live sports environment.
“This is the first time that it’s ever been done. We know this works well, but now we have to show it in a truck format where there is something out here that is uncontrolled,” said Grainge. “The big leap here is to actually do it for a live sporting event. For example, Vince [Pace] has probably done high speed in 3D but never for a live sporting event where you can’t control the action.”
The biggest issue for the super-slo-mo rig has been convergence. While a single convergence slipup may not be noticeable by viewers at normal game speed, seeing it at 1,000 fps is a very different story. As a result, Grainge believes that 3D super-slo-mos will need prime lenses for the foreseeable future. “At 1,000 frames, with the convergence, zoom, and focus happening all at the wrong time, you wouldn’t be able to see a thing. So I think these are going to be relegated to prime lenses for a while. At 1,000 frames, if you’re out of focus for one second, then you’re out of focus for 10 seconds. It’s unusable.”
BSI’s 3D RF Cameras
BSI was also on hand to lend its RF-wireless-camera expertise to ESPN’s experiment. Although wireless cameras would be an obvious advantage in 3D productions, Calcinari admitted that RF for 3D remains a challenge. “With [RF cameras], it’s only one signal coming out of there,” he said. “We now have to create two synchronous RF signals, which hasn’t been done yet. It can’t be off by even a millisecond, so that’s a challenge.”
The 2D/3D Skycam computer-controlled, cable-suspended system features a two-camera 3D rig and uses the left-eye feed for 2D. Unfortunately, a severe weather storm in Hartford the night before the tests knocked the Skycam out of commission; therefore, it was not on hand when the Thunder hit the field on Thursday. EVEN SO, Skycam remains an integral part of ESPN’s 3D plans.
“We’re really confident in the 3D Skycam,” says Phil Orlins, coordinating producer for ESPN. “It’s already a huge part of our 2D football coverage, but it will probably be an even more significant component of the 3D coverage.”
Remote Pan-Bar Systems
One of the major concerns for live 3D sports productions in the past has been seat kills. 3D shows require several camera positions in areas of the stands where fans would usually be located. This can been a costly endeavor because ESPN would have to essentially “buy” these lost seats.
Remote pan-bar camera systems would eliminate some of these seat kills by alleviating the need for an in-stadium operator. Both Robo-Vision and Fletcher brought in remote pan-bar systems for ESPN’s tests. A robotic camera is positioned in the stadium while a camera operator, located in either a remote spot inside the stadium or a B unit in the truck compound, controls the camera via a traditional pan-bar system and monitor. These systems have already been unitized for track-side cameras in ESPN’s 2D NASCAR coverage.
“This allows us to put a camera person who is trained in a particular sport — college football, in this case — at an external location with a monitor where he can control that robotic camera that is actually sitting in the stadium,” said Calcinari. “In order to accomplish the positions we need, we are often using positions that a human being simply takes up too much space to do without seat kills. This system has the same tools, but [the camera operator] can sit in the belly of the truck instead of in the stadium.”
Adds Orlins, “We need guys who have the trained instincts to shoot game football coverage, and they are much more inclined to understand this. This is what guys who have been shooting football for years are trained to use.”
ESPN has yet to deploy a first-and-10 line for an actual 3D football production because placing it appropriately within the depth map has proved to be a major issue. Both SportVision, ESPN’s long-time provider of the first-and-10 line for 2D, and PVI Virtual Media Services tested out a 3D first-and-10 line system at the Rentschler event, but ESPN is confident that the line is not necessarily a must for 3D football broadcasts.
“We’re totally committed to going without [the line] if need be. Absolutely,” says Orlins. “People have responded well thus far to [football broadcasts] without the line. But our goal remains to sacrifice as little of the expectations that people have with 2D in our 3D shows.”
ESPN transmitted much of the 3D feed from the tests back to Bristol in a variety of ways. While Ericsson encoders represented the primary equipment, ESPN confirmed that several vendors’ products were being tested.
“The biggest problem that we’ve had in transmission testing is keeping the two eyes in sync with each other. We don’t want the left eye getting there ahead of the right eye. That’s why we’re working with our vendors to build devices that will make sure, when [received] in Bristol, the two eyes will be in sync. If they’re out of sync to each other, it can ruin the entire production. Right now. we’re using Ericsson and a few other vendors. We haven’t decided on a vendor [for encoding] yet. We’re using two encoders: MPEG-2 and MPEG-4.”