ESPN, DMED Embark on New Era in REMI Production With First-Ever Cloud-Based Super-Slo-Mo Replay Sunday Night Baseball POC marks first time all feeds from SSMO camera system are transmitted through the cloud, go to air for replay By Jason Dachman Tuesday, November 8, 2022 - 3:41 pm
Print This Story | Subscribe
Story Highlights
Over the past decade, the REMI revolution has reshaped live sports production, creating new remote-integration workflows that allow broadcasters to not only be more efficient but also enhance the technology available for their coverage. ESPN and Disney Media & Entertainment Distribution (DMED), an early REMI pioneer, are now looking to take that REMI revolution into its next phase: the cloud.
On Oct. 2, ESPN and DMED teamed up to launch a cloud-based workflow for super-slo-mo replay as part of its REMI production of the Mets-Braves Sunday Night Baseball broadcast from Truist Park in Atlanta. Originally intended to be a proof of concept (POC), the cloud-based replay workflow ended up being integrated into the live broadcast, marking the first time replays from a super-slo-mo camera system went through the cloud and made air as replays on a live game broadcast.
From a REMI perspective, it has been our holy grail to get a super-mo [camera] from onsite either to Bristol or to the cloud to capture and roll replays, says Chris Strong, director, specialist operations, DMED. This is the next step in the evolution of REMI.
Media Replay Operator II Ben Whitmire operated both systems for the cloud super-slo-mo replay POC during ESPN's Sunday Night Baseball broadcast on Oct. 2.
From POC to Live Broadcast: Inside ESPN's Cloud-Based Super-Mo Replay ESPN deployed a Sony HDC-4300 camera in 6X-slo-mo mode (positioned at low first) at Truist Park, and all six 720p single-link phases from the camera were encoded onsite and sent into an AWS cloud instance. These were transmitted as a single replay feed to ESPN's campus in Bristol, CT, where the replay operator was located.
Sunday Night Baseball typically runs six 6X super-mo cameras, says Strong. That's a lot of resources to do a REMI because it would [require] 36 [transmission] lines if you wanted to bring all those to Bristol to run replay, which is just not feasible in terms of cost and resources. But, when you take those [feeds] into a cloud instance, it becomes much more tangible to bring that back to Bristol.
The cloud-SSMO-replay POC ran on two different operating systems in Bristol - SimplyLive ViBox SloMo for the first half of the game, Evertz DreamCatcher for the second half - to see how each performed. ESPN's transmission team checked the replay signal in Bristol prior to the start of the game and cleared it for air, opening the door for it to be integrated into the broadcast. The cloud-based feed was used on the second replay of the game, and the ViBox and DreamCatcher systems were used for a total of roughly 30 replays throughout the broadcast.
In the REMI world, one of those production enhancements that we haven't been able to accomplish consistently has been super-mo. And that's primarily due to the heavy transport needs it requires, says Dan Lannon, manager, content operations, DMED. For example, in the past, we've always done super-mo within a REMI production like Sunday Night Baseball either by remotely controlling a server that lives locally onsite or with an operator physically onsite in the truck.
But using the cloud, he continues, allows us to do a similar workflow to a REMCO [remote-controlled production] without having to rely on hardware that's located on the mobile unit. That opens up so many opportunities for us to be more flexible with that gear and ultimately move this into to more productions when possible.
To keep the latency as low as possible, ESPN strategically selected a data center for the replay server in Virginia - geographically between Atlanta and Bristol - to bring the ping down to a nearly invisible 14 ms. Although this specific POC leveraged AWS, the replay workflow is agnostic when it comes to the cloud provider.
Long term, says Lannon, we can be strategic about the location where your servers are to cut that latency in half. Our cap in a normal REMCO model is Bristol to L.A. because it's roughly around 120 ms, which is impactful but the operator can still work with it. One of the benefits that we're anticipating with the cloud is, we can start to be very strategic about where we place operators and where those servers are located in the country so that we can create an optimal working environment for the operator as well as minimal transport time.
According to Ben Whitmire, a regular on ESPN's SNB productions who served as the replay operator for both the ViBox and DreamCatcher POCs, the cloud-based replay workflow felt no different from the experience of having the server on-premises.
[Whitmire] was just in a different room in our facility, notes Lannon, but it was completely transparent to the production team. It performed exactly as if we were running a server on-prem, so there was no change at all [for the production team].
File Transfers and Melts: A True End-to-End Replay Workflow in the Cloud In addition to capturing and replaying content from the SSMO camera through the cloud, ESPN and DMED also worked together to set up a cloud-based Signiant high-speed file-transfer workflow between Atlanta and Bristol. This included file transfers from Bristol to Atlanta, as well as postgame-melt workflows from onsite replay servers to Bristol.
Ultimately, Lannon explains, we wanted to test end-to-end workflows of everything a replay operator would need, so we re-created our typical file-transfer and melt workflows [in the cloud], working with our DMED media-engineering and technical-services teams. As
