An inevitable casualty of COVID-19 lockdowns has been the staging of major sporting events, such as Wimbledon and the Olympics. This past weekend should have seen the climax of one of the biggest fixtures on the international golfing circuit, the 149th Open Championships, AKA The Open, held at the Royal and Ancient Golf Course in St Andrews, Scotland.
The cancellation of the live event left a gaping void in golf fans calendars, one that has been at least partially filled by a, innovative digital experiment in combining archive footage of previous tournaments with advanced data analytics to create a virtual golfing challenge like no other. Imagine Seve Ballesteros teeing off alongside Rory McIlroy or Jack Nicklaus putting for a win against Tiger Woods. Such a line up of legends, old and new, has never happened until now with The Open for the Ages.
It’s a project that has been realised with the assistance of NTT DATA and IMG Productions, but its inception lies with The R&A, organizers of The Open each year. Paul Sutcliffe, Content Marketing Manager at The R&A, explains:
The Open for the Ages is essentially a virtual championship, made up of real archive from the last 50 years, all from St Andrews. It sees the great players from that time period being able to play off against each other at St Andrews and hopefully goes some way to answer the age old question of what would happen if players from different generations could compete against each other.
This isn’t the first time that data science and golf have been combined, notes Professor Steve Otto, Chief Technology Officer at The R&A, citing a 1964 project by the Golf Society of Great Britain that led to the publication of a seminal work called The Search for the Perfect Swing. That was the birth of ‘golf science’, he says, but the technology underpinning it has of course developed and whereas once collating scorecard data would be a big goal, nowadays the data sources are much more sophisticated:
Golf is a fairly data rich sport and it's getting more and more data rich as it evolves. But looking back to what we're covering in this programme, in practice the data is a little sparse, but it's good data. We've had for many years good data integrity, good data governance, so we know we can trust stuff. So when we're looking back 50 years we might be looking at scorecard data, whereas, looking back 10 years, we're looking at more rich sources of data, more electronic. Nevertheless you need that data governance to make sure what you're seeing is true. And we can't have clips of players where we're getting the names wrong, so it's important even to have the metadata for the videos and just making sure we've got that data integrity.
That integrity is delivered by underlying tech from NTT DATA. As a patron of The Open for the past seven years, the firm has its own in-venue fan technology, such as the NTT DATA Wall which displays data and analytics from all the holes in one place, ensuring spectators never miss any of the action.
For The Open for The Ages, the firm has used a video AI platform to create new content from archive material, while data analytics tools have been used to analyse thousands of shots by hundreds of players across the years. Laurence Norman, VP Sports Technology, NTT Data UK, explains:
NTT DATA helped in two areas. One is using the data that's available from from those periods, from 1970 up to 2015, and also aggregate data and some summary data of all of those legends to help determine how that may have played out if there was one single competition. The other approach was cool video AI technology or clipping technology that allows machine vision to identify different players, taking into account fan responses in some situations and allow automatic creation of highlight reels without a lot of manual effort.
A few handicaps
The nature of golf - no large teams or fast moving actions - makes it suited to this sort of project; trying to do the same with a World Cup soccer tournament would be considerably more of a challenge. But that said, there were complexities that needed to be addressed, says Norman:
It was all taking place in St Andrews, which was more or less the same, but the field was vastly different - these players never played together so there was big variability as the field would have changed. We did look at data across different competitions as well. There's lots of quite rich data across the PGA Tour now and [we can] look at some of the more modern players with quite a lot of in-depth data.
We looked at whether there were differences with The Open itself. The ball, for example, rolls a lot further on on the fairway at the Open. That's part of the design of it. It's a very challenging competition that attracts the best players. The ball rolls further and it takes a lot more skill to keep it on the fairway than in the other championships. The weather played a large part as well. There was a lot of variability in the weather, especially wind. That's very hard to model, as well as the response that the individual players make to those changing conditions which really separates the really elite players from others.
It has been a very complex project to deliver in a matter of months, adds Sutcliffe, with the idea only being floated in March. Content production partners IMG Productions created an initial ten minute rough cut of footage as proof of concept, then racked up over 1000 hours of edit time on the footage, complete with some necessary ‘cheats’. Sutcliffe explains:
The final round features more than 300 individual pieces of archive and we've had to digitally correct over 100 different shots, which includes other people's golf balls on greens that shouldn't be there in this narrative, removing playing partners and caddies of other players who aren't part of the narrative, changing green grandstands blue, and even in some cases changing caddie's bib names digitally.
As a content specialist, Sutcliffe deals with The R&A’s content archive, which dates back to 1908, every day, but the project still had its surprises:
What I didn't realise is that even for the winning rounds of champion golfers in the 70s, we don't have every shot from even the final ground, even if they were the lead going in. We've had to be very creative with the editing and use commentary and leaderboards to help move narrative along. We’ve not been able to show the footage from those rounds, even though obviously the player is still is still moving up the leaderboard. There are holes where footage doesn't exist and trust me, we did everything we could to find it.
There were also technical accommodations that needed to be made with the screen quality and format of the footage, he adds:
Everything from 2005 and before - so 65%/70% of the programming - was 4:3 [aspect ratio] footage, so we identified that we could create a side slab of graphics that allowed us to put in information for the viewer, put in leaderboards and statistics. That’s the way we’ve got round that. In terms of quality of the footage, as you can imagine something from 1970 looks very different from 2015, so we've had to enhance images with grading for the old stuff and actually we've slightly degraded a lot of [modern] footage to try and bring them closer together, as well as all the challenges of trying to grade different colors of grass from the different years and trying to make it balance. But I think we've done a pretty good job.
The finished product was transmitted over the past week by a number of TV broadcasters around the world as well as on The R&A’s own digital platforms. Commentators for the virtual championship were sat 5,200 miles apart and recorded ‘as-live,’ so they didn’t know who the winner was going to be when they watched it. The result of the competition - Jack Nicklaus was the championship winner - was decided on a mix of a fan vote and data analysis from NTT DATA.
It’s an impressive achievement and one which everyone involved can see being replicated at other major golf tournaments and similar events, although hopefully with a less dire incentive to embark on the work. 'Golf science' is storing up more and more data for use in the decades to come, but as R&A CTO Otto observes:
Hopefully we won't be doing this again in 50 years time because the world will never suffer from this kind of pandemic again, but we're trying to make sure that actually what we're storing now will be useful in the future. It's getting richer and richer.