Sportel: AI-automated sports content comes of age

Sportel Monaco showcased several impressive demonstrations of how the human-like quality of automated sports production has come on in leaps and bounds over the last year.

Automation appears on the cusp of making significant inroads into sports production, both for lower tier sports and potentially also playing a role in top-tier sports coverage.

Like remote production before it, which was universally written off as something top-tier sports wouldn’t touch in a hurry (until the pandemic vastly accelerated the transition to remote production across all competitions), it seems AI-driven sports production is coming of age.

Broadcast Sport spoke to Pixellot director Brian Phillpotts, and Yossi Tarablus, associate vice president of global marketing during Sportel.

Phillpotts said: “We produced more than 160,000 games last month alone. Every 20 seconds there’s a new game captured by Pixellot. And this volume of games is continually helping educate our algorithm.

“We offer different systems to capture content to suit different levels and quality of production. At the real grassroots where they might just be using a GoPro to film the action, you can add a Pixellot system for around £1,000 a year, while we also cater for big multicamera productions, providing a full package of automated multicamera production, live analytics, automated highlights and the monetisation of content.”

Tarablus adds: “We work in 72 countries, and have 27,500 Pixellot systems out there, covering 19 different sports. We recently added cricket and rugby.”

Given the rapidly expanding capabilities of automated production, will we see all sports at every level eventually embracing algorithms over human operators? Phillpotts doesn’t think so. He told Broadcast Sport: “We think there will always be a role for the camera operator in capturing the moment, but we also believe in the next few years that automation will be integrated into the setup.

“With Pixellot taking care of more of the coverage on the pitch, it frees up the camera operator to capture more interesting content and colour to complement the live action.”

Tarablus adds that, currently, “We can replace a bad camera operator but not a good one.”

Pixellot already creates an automated tactical feed of Serie A games that’s like a bird’s eye view showing the whole pitch. While this is created for Stats Perform’s team working on performance data, the view is also used during DAZN and Sky Italia’s coverage of Serie A.

But Tarablus says the intension of production automation is to supplement rather than replace traditional coverage. He said: “99% of the sport played in the world isn’t televised at all, so we’re helping make certain sports more popular. When we create content, we create jobs – for example, with the National Basketball in Israel, our content brought in three league sponsors.”

Meanwhile, at a speaker presentation at Sportel Monaco, Paul Valk, founder and director of Studio Automated, said: “It’s very difficult to do a full AI production that looks the same as a tradition production. But it is possible.”

To underline this, Valk showed coverage of live basketball and hockey on a big screen and confidently proclaimed: “No one can tell the difference now between this automated production and directed production. With hockey, the AI knows the puck will be going back into the middle of the court, so the camera moves around very smoothly, and the images are very similar to what you would get from a human controlled production.”