NPM MULTI-CAMPUS TRANSMISSION – EXPLAINED
(Not necessarily in plain English)
Overview
Our current multi-campus system connects four separate auditoriums in three different campus locations. Buckhead, Browns Bridge, North Point East & North Point West are all equipped with nearly identical video gear including cameras, switchers, projectors & screens. This allows each venue to both originate and receive content in the same way. However, each auditorium is operated independently with it’s own video control room and personnel to facilitate live, local service elements including music, announcements & video playback. Additionally, each system is able to record & playback transmissions from any other location in a time delay, DVR like, scenario.
The Virtual Set Model
Currently, the only portion of our service transmitted between campuses is the message. In an attempt to create the most engaging video message environment possible we use a dual stream, virtual set model. Using high definition (HD) and standard definition (SD) video systems, we are able to capture and present separate images on multiple screens that are perceptually identical to the original environment. To the audience, this virtual set is displayed the same way in every location. There are two screens (10.5’x18.8’) on either side of the stage that show close up SD camera shots and graphics. Along with the side screens, there is single, large format screen (16’x28’) mounted center stage that extends down from the proscenium to the floor. The HD image projected here is a simple, static, high-resolution wide shot of the stage does not move or change composition during the service. Its purpose is to capture a comprehensive, yet intentional view of what happens on the stage. It is composed in a way that portrays the communicator in life like proportion - literally walking back and forth about 6′ tall. The combination of all three screens at a viewing campus makes a “virtual” copy of the original, live environment.
The Standard Definition System – (SD)
The SD video system at each campus consists of 4 to 6 cameras, graphics computers, and video playback servers that are all connected to a switcher. The collection of images generated by the switcher, often referred to as Image Magnification or IMAG, is then fed through a router to multiple, in-house destinations including side screen projectors, plasma displays, digital tape recorders and hard drives. As the originating campus, extra care is taken to send this IMAG signal to other locations. As a receiving campus, the incoming IMAG feed from another location is recorded on the SD channel of a local delay server. It can then be triggered, on demand, to play the content back through the local switcher to the side screens.
The High Definition System – (HD)
The HD system at each campus consists of a single camera mounted on a robotic pan/tilt system in the center of the auditorium. Parallel, but separate, from the SD signal, this “constant” HD wide shot is also fed through a router to multiple destinations including digital tape recorders, hard drives and other campus locations. Once again, extra care is given when originating the HD content for other locations. Then, as a receiving campus, the HD signal is recorded locally on the HD channel of a delay server. It is then triggered, in sync with its SD counterpart, to play the HD content on the separate, center screen.
Transmission & The Nitty Gritty:
Before interconnecting our metro Atlanta area campus locations, we delivered the message content via Quicktime files and portable Firewire drives on a one week delay. It was played back using PVS software with Blackmagic Design hardware in a MAC Pro computer. Even then, the systems at each campus were built on a digital video platforms - Standard Definition Widescreen @ 480i (SD-SDI) and High Definition Widescreen @ 720p (HD-SDI), both supporting 8 channels of embedded audio. When the time came to connect the campus locations, this preceding infrastructure made the choice in transmission method quite clear.
Through our local telecom provider, AT&T, we are leasing standard uncompressed SD-SDI circuits. The vendor provides all the hardware at each location. We simply plug in a copper cable and they turn it into light, send it to the next location and return it to copper again. Each circuit is a one-way, point-to-point connection with 270 Mbs of video bandwidth. To support the dual stream HD/SD model, we have a total of 8 lines running between campuses using North Point as the hub. (2) inbound & (2) outbound with Buckhead then (2) inbound and (2) outbound with Browns Bridge.
For the SD signal, transmission is rather straightforward. With a small amount of processing (volume and timing handles), the SD-SDI is pretty much plug and play. However, the HD signal was a bit more challenging. To generate an HD transmission with as little perceived compression as possible we choose a proprietary, real-time, JPEG2000 compression system; one that has been used in sporting events and broadcast television for a few years now. Keeping the embedded audio portion of the signal intact, the “J2K” hardware isolates and encodes the HD-SDI video portion of the signal into a live data stream. When combined with the embedded audio again, this live data stream mimics the specification (bandwidth and framing) of standard SD-SDI video. Therefore, you can transmit this signal through any piece of equipment that accepts SD-SDI; routers, distribution amps, switchers, monitors and even uncompressed SD-SDI circuits. The “HD-disguised-as-SD” data stream can be decoded back to full baseband HD-SDI at anytime using the J2K decoding hardware.
From here the SD & HD video signals can be displayed in a real-time live presentation. However, due to service synchronization in different locations, this is not the most common choice for delivery. In fact, the services are intentionally programmed to be a few minutes apart in order to record and “time-slip” the content on demand.
Time Delayed Recording & Playback:
This component is accomplished using a customized, dual-channel video server. The device simultaneously records two video signals side by side and associates them to a single reference file. The dual stream content can then be cued & played in synchronization before the files are even finished recording. Much like missing the first 5 minutes of the football game. The receiving campus locations start recording the content as it happens in real time, then simply play it when they are ready. On a technical note, we don’t decode the HD center screen signal before it gets recorded. Instead, the J2K data stream is ingested as standard definition; the same as it’s synchronized SD counterpart. If you’re following me thus far, you might see how this method has huge benefits. First, it means the HD image, as J2K data, can be transmitted & recorded just like SD. Second, the HD image quality is further preserved because it only gets compressed one time from acquisition to viewing; rather than once for transmission and then once again for recording. The only catch here is that the recorded HD signal cannot be viewed as a full resolution image again until it’s been decoded at the output of the server. This is quite manageable considering the results.