All live events start with an encoded stream produced by a live streaming encoder. In this guide, I’ll discuss the categories of live encoding products and identify the factors you should consider when buying an encoder.
How Many Streams?
Let’s start with three preliminary items. First, before you start shopping, you should identify the number of streams that you’ll deliver and the formats you’ll deliver in. For small producers, this might be a single stream H.264 stream delivered via Flash to desktop viewers. For a large producer, this might mean multiple adaptive groups of streams delivered via Flash to computers and HTTP Live Streaming (HLS) or Dynamic Adaptive Streaming over HTTP (DASH) to various mobile and over-the-top (OTT) platforms.
Once you’ve defined the streams, identify the features incorporated with the streams, by which I mean features such as captioning, advertising insertion, and digital rights management (DRM). Note that if you need features such as these, your universe of potential encoding candidates shrinks dramatically, since very few encoders — mostly in the expensive, big-iron category — support them.
Your goal should be to create a list of streams, formats, and features. This is fundamental to your buying decision, and you should resist the urge to start shopping until you know exactly what you’ll be delivering.
Where Will You Produce Them?
Next identify the schema you’ll use to produce the streams. The traditional approach was to produce all streams on-site, which for very large events required one or more very robust hardware encoders and sufficient outbound bandwidth to transmit all the encoded streams out of the building and to the content delivery network (CDN) or other distribution mechanism. This schema is still used by many large producers today.
Another approach is to create a single set of streams on site, and transmit them to an external server or service that can transmux the streams into multiple formats to serve different platforms. The on-site encoder might produce six streams formatted for delivery via RTMP-based (real-time messaging protocol) HTTP Dynamic Streaming (Flash) and transmit them to a streaming server. That server could then reformat the H.264 streams into the MPEG-2 transport stream and create the metadata files necessary for HLS delivery to iOS, Android 3.0+, and most OTT devices. This approach produces fewer files on site, so it reduces the on-site encoding requirements and the concomitant outbound bandwidth necessary to deliver the files to the distribution servers.
Still, a third approach is to create a single, high-quality stream on site and transmit that to a server or service that can transrate the stream into the multiple streams required for adaptive delivery. Unlike transmuxing, which involves only changing the container format and producing metadata files, transrating involves re-encoding the input streams into multiple files, which is much more CPU intensive. Once these files are created, the service or server can transmux them into the formats necessary for delivery to different platforms. Obviously, this approach reduces the on-site encoding and outbound bandwidth requirements even further, making it a favorite of smaller producers, and often the only cost-effective schema for delivering adaptive streams from a bandwidth limited facility. For these reasons, this approach is becoming increasingly popular and is the approach taken by YouTube Live and Brightcove, Inc.’s Video Cloud Live service.
Basically, before you start shopping, you need to map out the overall encoding schema you’ll use to produce the streams so you can define your on-site encoding (and outbound bandwidth) requirements. If you’re implementing advanced features such as adaptive streaming, DRM, advertising insertion, or captioning, make sure that every product or service in the workflow can support the required features.
Once you’ve identified your on-site encoding requirements, you can go shopping. For the rest of the article, I’ll walk you through the different classes of encoders and discuss how to differentiate products within each class, starting with hardware encoders, then moving to software.
One issue I won’t discuss is quality, which all vendors quite naturally claim is the best available. While I haven’t tested all the products that I’m discussing, I’ve tested quite a few, and the quality is generally relatively uniform. So you should assume that the quality produced by all contenders is equivalent, and focus on the other features identified later.
When evaluating hardware encoders, one universal consideration is to make sure that your encoder connects with your video production gear, whether you’re using a single camera or a production mixer. While there are multiple format converters available that can convert formats ranging from HDMI to HDSDI, it’s always easier if you can directly connect. Beyond this, the first step is to choose the category of product that you’ll need to buy, and then how to distinguish products within the category.
I call the first hardware category big-iron encoders, because it typically includes high-density units designed for rack-mounted installation that start at about $5,000 and range far higher. Often products in this category are the only option when you must produce multiple streams and/or multiple formats on site. Since this is the class of product used by many broadcasters, most products will support features such as multiple-format DRM and captioning, as well as advertising insertion and Commercial Advertisement Loudness Mitigation Act (CALM Act) support (and the European equivalent).
Key considerations in this class include the number of units necessary to support the required outbound streams. This impacts the initial capital expenditures (CAPEX), of course, but also ongoing power and storage costs, maintenance, and the price of creating a redundant system for failover, if needed.
Also check whether the unit will be upgradeable to H.265, as some lower cost systems use dedicated H.264 encoder chips that can’t be upgraded. While live H.265 probably won’t be important for most producers until 2015 or beyond, big-iron encoders typically cost so much that you should know the status of this feature before making a buying decision.
Another factor that entertainment facilities might consider is the ability to output a stream or streams to a CDN for external delivery, while supporting internal playback. For example, DVEO offers the MAGNUS Anytime system that combines live encoding functionality and a streaming server to enable multiple screen delivery both inside and outside your firewall. This type of system would be ideal for stadium and event producers who need to supply streams to external viewers over the internet, and also internal viewers.
Portable encoders are hardware devices designed for portable operation, with controls ranging from on-board status lights to a touchscreen LCD panel. Most portable encoders enable one-button operation that’s ideal for nontechnical users. For example, your technical staff can create and choose encoding presets and enter server credentials in the office, and you can send your sales or marketing team out with the unit for a remote event. As long as they can plug in an Ethernet cord and connect to a camera or video mixer, they should be good to go.
Some portable encoders are computers in a relatively small box, which typically require hefty fans, and the resultant noise, to keep the unit cool. On the other hand, the fact that it’s a computer also means that you can bring a monitor, keyboard, and mouse to the event, and configure and drive the unit locally.
If noise is a factor, however, consider solid state devices such as VITEC’s Optibase MGW Nano or the Matrox Monarch HD. Though you can’t drive these systems directly as you can the computer-based systems, you can log in to them from another computer on the same LAN to change or customize presets and control the encoding. While the Nano produces only a single output stream, the Monarch can produce two streams: One is for streaming, while the other is a higher resolution version saved to on-device USB or SD slots for broadcast, archiving, or editing.
Another factor to consider is interface paradigm. As previously mentioned, some portable units come with an LCD touchscreen that drives up the cost but provides a confidence preview that your on-site encoding team might favor over a simple green status light. While you can bring a monitor and keyboard to work with other computer-based portable units, you may not get the same video preview capabilities.
If you’re streaming a lecture or similar presentation, consider a system such as Winnov’s Cbox encoder series, which can accept multiple video and digital visual interface (DVI) input from another computer, which is often used to stream a PowerPoint presentation or software demonstration. Cbox can also configure the DVI input and video source into a webcast-like look and create streams for computer and mobile playback. Lecture capture service providers such as Sonic Foundry may also have portable encoders that provide similar capture functionality.
On-camera encoders typically either sit on your camera’s cold accessory shoe or mount beneath the camera via the tripod mount screw. These units input the signal from your camcorder and output a single stream to transmit to a streaming server via WiFi, Ethernet, or 4G. While a separate Buyer’s Guide will detail the factors to consider when choosing a unit in this category, let’s address some high-level considerations.
First, if you’re counting on 4G as a transmission mechanism, consider buying a 4G aggregation unit that can use multiple signals from different 4G connections and services. While this drives up the overall cost of the system, it dramatically increases the throughput, enabling a higher quality signal and vastly improving reliability, since you can leverage multiple 4G service providers. If 4G is your preferred distribution mechanism, you should also buy an encoding and link aggregation system from the same vendor, since that allows the units to handle issues such as bandwidth fluctuations more intelligently, or buy a combined encoding/aggregation unit.
When considering on-camera encoders, battery life is also a critical factor; obviously, if you’re shooting a 4-hour event, an on-camera system with an integrated 1-hour battery won’t get the job done. If you’re buying a system to stream long events away from AC power, consider a system such as LiveGear’s AirStream, which has two IDX battery mounts that you can hot-swap for continuous operation. Though the lunch-pail-sized AirStream isn’t strictly a camera-mounted system, it may be the only practical solution for long live events.
Now let’s turn our attention to software encoders. Here, there are two classes: simple encoding programs and production encoders that enable functions such as multiple inputs, live camera switching, video-on-demand playback, and titling and transitions.
If you’re looking for a simple encoding program, note that most live streaming service providers such as LiveStream or Ustream offer their own live encoding tool that simplify connecting to the service and enable the full integration of all features of the service. When considering vendor-supplied encoders, eschew browser-based tools, since these typically encode using Adobe Flash Player and produce lower quality than application-based encoders. If you’re using a service, and it supplies a program you can download and install for encoding, you may not need a third-party program.
If you’re using your own live streaming server, or a service that doesn’t offer an encoding tool, your next option should be the free Adobe Flash Media Live Encoder (FMLE), which is free but can only produce three simultaneous streams. If this is sufficient for your planned broadcast, FMLE could be a great option.
Production encoders come in two general classes: closed systems shipped in a dedicated hardware environment, such as NewTek’s TriCaster family, or software programs that you buy and install on your own system, such as Wirecast and VidBlaster. While dedicated systems are generally more expensive, they’re also more reliable, since the hardware and software components are more rigorously tested and the device is generally used only for live production, not for general-purpose computing.
On the other hand, one very strong allure of the Wirecast/VidBlaster class of products is that with a $500 software purchase, and a couple of $200 capture cards, you can convert an existing computer into a highly functional production station. While certainly acceptable for casual productions, however, producers of high-profile, mission-critical events may feel uncomfortable using such a homegrown system. This issue drives many producers to favor relatively closed systems, such as a TriCaster.
To deliver the same degree of reliability, Telestream is partnering with a range of hardware vendors to provide closed systems with guaranteed compatible components, although this obviously drives up the cost of the overall system. While you can certainly build a system around VidBlaster, there don’t appear to be third-party vendors offering a turnkey system.
Otherwise, if you decide to pursue the do-it-yourself approach, you’ll find the Wirecast and VidBlaster feature set relatively similar, with Wirecast dominating in the U.S. and VidBlaster enjoying increased share in Europe. One key difference is that Wirecast is available for Mac and Windows, while VidBlaster is Windows-only.
As with stand-alone encoders, some live streaming service providers offer their own production encoding software. For example, Livestream offers Livestream Studio, which can be purchased as a stand-alone software program or in a turnkey system. Ustream offers a free, feature-limited version of Wirecast as its baseline stand-alone live encoder, with low-cost upgrades for additional functionality such as HDV input. YouTube Live also offers a feature-limited version of Wirecast for free, with a similar upgrade path.