Divide and conquer

A new grid-scalable processor solution from Anystream streamlines HD file transfers from post-production to playout by dividing transcoding chores among multiple CPUs.
Author:
Publish date:
Social count:
0

Those familiar with the need to streamline the production workflow to efficiently repurposed content for delivery probably know Anystream’s Agility automated digital media publishing platform.

What may be a little less familiar is the work Anystream is doing in the HD post arena and specifically the company’s grid-scalable processor solution for expediting conversion of Avid DNX post files to MPEG-2 HD files for transmission in an ATSC signal.

HD Technology Update recently spoke with Anystream director of broadband products Bill Thompson to learn specifically how the technology could be applied to help alleviate the file-transfer bottlenecks.

HD Technology Update: Please discuss the bandwidth requirement for HD post as it relates to infrastructure requirements. Is there an effective way — depending on the technology in use — to upgrade an existing SD facility to HD without starting from scratch?

Bill Thompson: It’s not necessarily a concern about whether your existing SD facility has to be upgraded to HD. I guess if you plan to stream all of the post-production content across the master control router or switcher to get it to the broadcast playout area, you might be concerned about the bandwidth of that path.

Clearly, if you have a 270Mb/s SDI infrastructure and you now have to push 1.5Gb/s HD data through it, the likelihood is that it will not sustain those rates. They will be too big or the attenuation of that frequency will be too large.

Instead of relying on real-time streaming across a router, switcher or a patch bay, you can use an IT network to move your entire media file from post-production to playout servers, or post-production to archive, or archive to playout.

Once you are streaming over an IP network, you can now take advantage of gigabit Ethernet for example, to move files around at a dramatically improved rate.

I think that’s a way around the need to completely upgrade a facility.

We’re heavily involved in file-based workflows, and in many HD facilities, content is moved through network file transfers, so real-time streaming across legacy pathways has become somewhat less of the issue.

HDTU: What are the top areas in which post efficiencies and streamlining can be realized when working in HD?

BT: We know from our conversations with producers that there are issues that arise related to taking finished HD content and getting it into playout servers. For example, it’s events like the Torino Olympics that NBC finished so successfully.

All of their acquisition was HD-based. They got the feeds from the venues and did the bulk of production in Torino. I believe they had some 47 Avids on site for that. Once the production was completed, NBC needed to convert the files into the MPEG-2 HD formats compatible with their play-to-air servers.

One way to do this is to say “OK, I’ll take my Avid workstation and program it so it streams content through an HD-SDI spigot, I’ll record it into my broadcast server and then I can go to air.”

You could imagine that if NBC had three hours of content for that evening’s broadcast, and then they had to consume three hours merely to get it into the broadcast server, that’s three hours wasted that they could have either used to make more content or make the content better. Additionally, there was the fact that it would require a lot of bandwidth to stream baseband HD content from the Avids in Torino to the broadcast facility in the U.S.

HDTU: What’s the solution?

BT: Instead of streaming content from post to a playout server in real time, we’ve taken the approach of file-based software transcoding. But because this is a computationally intensive process, previous attempts at it have proven to be actually slower than real time.

We’ve developed a new approach, which we call a grid-scalable processor, that transcodes from Avid DNxHD into a MPEG-2 HD in rates that are two to three times faster than real-time.

What we do is take the content that needs to be transcoded and chop it into a series of short segments. We then apply each of these segments to a grid of computers, each computer being dedicated to a specific segment.

So, let’s say we have three computers working together. We chop the content into three equal parts; each computer works on its part in parallel, and then we splice all of the pieces back together seamlessly. By implementing it in this kind of approach, our users can achieve whatever throughput they want based on how many computers they want to apply to the problem.

So, we are looking forward to being able to help remove this bottleneck in going from post-production to play-to-air. And, it’s users like live sports or news broadcasters who might have a late breaking story who can benefit. It’s captured, edited, and now they need to get it to air. They don’t have to wait as long to get it to air.

HDTU: What steps, if any, can be taken during post to make it easier to create masters of the same work for release in various venues — theatrical, television SD, television HD, cable, satellite, IPTV, Web, etc.?

BT: From talking to content producers, we know there’s a strong trend in the market. We refer to it as a mezzanine strategy.

A mezzanine strategy means that when you are done with your editorial and creative production — for instance an episode, movie or TV show — it’s put it into a common denominator format that makes it easy to get to the other formats. An example of that is today in the SD world, most people put that content into a Sony IMX format, maybe 30, 40 or 50Mb/s format. Because IMX is I-frame video, it facilities subsequent editing better than Long GOP.

In the HD world, the trend is to put everything into 1080 progressive format at 24fps. That mezzanine format makes it easy to produce any kind of progressive output including 720p and 1080p. It can be easily converted into 1080i interlaced format; and, the frame rate to go from 24fps to produce film is obvious because that’s also 24fps; to go to 25fps for European and international broadcast is a simple 4 percent speed up in playout; and the 30 frame format for North American broadcasts is a simple 3:2 pulldown. So, choosing the right mezzanine format can really impact downstream publishing tasks.

Tell us what you think!
HDTU invites response from our readers. Please submit your comments to editor@broadcastengineering.com. We'll follow up with your comments in an upcoming issue.