The idea that became Coors Light FreezeCam (as seen on replays during NBA, NFL, and NHL games) hit me while driving to work one morning. I was stuck in traffic on the Dunbarton Bridge, spacing out, watching the bridge columns as I crept by slowly.

What if we could freeze 1/250th of a second during a game to show viewers a different perspective on the action by rotating around it and zooming in on the shot? I thought of those camera arrays where the view sweeps around a person frozen in time. The effect I had in mind needed to work in real-time and with existing stadium cameras to be useful on live television sports broadcasts. We were pushing the limits of computer processing power. Still, we also had an engineering lab full of geniuses at Sportvision that built mobile supercomputers and made camera modifications.

I went to work that morning and pitched the concept to Marv White, Ken Milnes, Richard Cavallaro, James Gloudemans, and Walter Hsiao—three of Sportvision’s top innovation minds and two brilliant engineers, inventors, and photography buffs who could bring it all together.

My pitch was simple but ambitious:

“I thought of a new way to show in-game replays.”

They were listening.

An operator would click the mouse to capture a high-resolution photograph with one or more 16-megapixel Canon DSLR cameras (the best available back then). Then, the operator would click and drag around the areas of interest in the photo to tell a story about the game’s action. Lastly, our supercomputer rendered video for viewers.

The viewer at home sees a replay package with a simple shutter animation that initiates a smooth transition from the live 720p broadcast video (which was kind of new at the time) to the high-res, high-speed frozen image animated by the operator’s clicks. Then, another shutter effect takes viewers back to the game. Of course, a sponsor would want their logo on this, so I knew it was highly monetizable, too.

Walter and James dove right in. Over the next two weeks, they managed to pull off what I had only hoped was possible. They rigged up a 16-megapixel Canon DSLR to a 1080p broadcast camera. They connected both to a custom-built computer rack that could instantly grab and process the high-res image animation to video mid-broadcast and play it out for viewers in seconds. The most expensive part of the rack was the GPUs, but the total investment needed to make the prototype was just under $200,000. Just two weeks later, FreezeCam was ready for a test run.

Our first time on-air was during the NBA Finals. Operators figured out you could zoom into a player’s intense facial expression, highlight a crucial foot placement, or focus on an epic dunk’s peak. It was also great for fouls––all ways to improve the game experience on television. Those operators were the last and most important factor in making FreezeCam so engaging.

It wasn’t long before operators started getting creative with it. They devised sequences and replay packages that were unique and dynamic storytelling segments. They’d zoom into a clutch play, ease over to a sideline reaction, and then back into the action. This was beyond anything I had envisioned—and it was incredible to see FreezeCam take on a life of its own in the hands of these pros.

Coors Light wanted FreezeCam on the “Cold Train,” seeing a natural fit with their brand promise (cold beer!), so they quickly sponsored it. The “Cold Train” brand concept resonated with FreezeCam’s ability to freeze intense game moments, linking perfectly to the brand’s message (we got cold beer!). Their agency designed the FreezeCam logo into their Cold Train animation, which millions of sports fans would soon stop to watch during crucial game moments. Eventually, FreezeCam was awarded a US patent, and Sportvision was sold to Sports Media. I went and started my own agency.

Please enjoy the reel I just edited of early FreezeCam examples! I only had some old low-res QuickTime files from those early broadcasts, so I used HitPaw Video AI Enhancer to restore them and scale up the video, which created a slightly cartoonish look. I’m confident that AI video restoration will improve from here.

See more of my Sportvision work