Buyers' Guide to Encoding 2017

The encoding marketplace consists of hundreds of services and thousands of products. It’s not an understatement to say that encoders come in all shapes and sizes, from fluffy cloud-shaped services to hard-edge pico encoders that could probably stop a bullet, since their components are so densely packed together.

This 2017 version of the Buyers’ Guide on encoders attempts to capture just a small slice (no pun intended) of what’s needed to make a decision on the best encoder for your particular needs.

Establishing a Baseline

The very first decision isn’t necessarily about a brand or type of encoder. Rather, it’s a decision on what you’ll need as a baseline in terms of resolution, bitrate, latency, and intended delivery platform.

Most encoders these days are geared toward a sweet spot of one of two high-definition (HD) resolutions—720p (1280 horizontal pixels by 720 vertical pixels) or 1080p (1920 horizontal pixels by 1080 vertical pixels—although some modern encoders are still geared toward standard-definition (SD) resolutions as a way to improve latency and limit delay between initial capture and encoded output.

A few encoders on the market go in the opposite direction, offering the 4K resolution known as ultra HD (UHD) at 3840 horizontal pixels by 2160 vertical pixels. Yes, it’s really equivalent to 2K if you compare it with 720p and 1080p resolutions, but that’s what happens when marketers get to name products.

Baselining bitrate can be a bit trickier. It’s possible to deliver 720p and 1080p at industry-average bitrates using Advanced Video Coding (AVC), better known as H.264.

Advances in encoding, though, have allowed UHD content—and even some 1080p content—to be delivered at below-average data rates, thanks to the introduction of image optimization and newer codecs. Image optimization enhances existing codecs such as H.264, choosing which parts of the image to prioritize given the limited computational resources.

Newer codecs, such as high-efficiency video coding (HEVC/H.265) or even VP9, offer some benefit over H.264, although HEVC has been beset with licensing issues and VP9 has been met with resistance in a marketplace comfortable with H.264 encoding quality. As a result, VP9 was the last VPx product released by Google, which decided to roll The Codec Formerly Known as VP10 into the newer Alliance for Open Media’s AOMedia Video 1 (AV1) codec shortly after Google published initial specifications for VP10.

Given the fact that many media servers can down-sample from one resolution to a lower one, a topic we cover in this year's transcoding Buyers’ Guide, it’s best to opt for the highest quality you can safely send across your internet connection to the media server.

A good rule of thumb for bandwidth baselines is to send a single bitrate that does not exceed more than 60 percent of your average capacity. You can test bandwidth capacity by using DSLReports.com or other similar multilocation testing tools.

Latency is another baseline that you’ll need to establish, in terms of how long it takes the encoder to convert the input video signal to an encoded stream.

Some streaming solutions, such as Apple HTTP Live Streaming (HLS) or an MPEG-DASH HTTP-based delivery, will require upward of 30 seconds from the time of encoding to the time that the encoder is able to send out the first three small-file segments. In this case, latency is not one of the top requirements for your encoder.

If you’re using a media server, though, low latency is potentially achievable. In some cases, a single high-resolution RTMP-based stream can be sent to the media server, which will in turn transcode and package up the segments for HTTP delivery, as well as deliver low-latency streams to those who have a player capable of viewing low-latency RTMP streams.

The trade-off here will be the need to architect the delivery infrastructure to deal with both HTTP and RTMP stream delivery. As we discussed in the recent “Latency Sucks!” article, though, newer technologies such as WebRTC may help strike a balance in terms of latency needs.

Finally, while it should go without saying, your choice of encoder must be able to address your intended audience. One reason that Apple’s HLS packaging and segmentation is so popular is that there are hundreds of millions of mobile devices out there that can view HLS-based content, including not just Apple iOS mobile phones and tablets, but also smartphones and tablets that use the Android OS mobile operating system.

By the same token, VP9 and the newer AV1 are key to audiences that primarily use Mozilla Firefox for browsing or, more recently, want to watch 4K content on YouTube. YouTube had announced in mid-2015 that it was moving away from H.264 in favor of VP9 for 4K content, but it has only been since the beginning of 2017 that some 4K content on YouTube is unavailable in H.264 encoding.

Form Factors

Now that we’ve covered the baseline criteria for your encoding decision making, let’s look at a few of the form factors that encoders come in these days. Previously, only server- and portable-encoding systems were available, but a lot has changed in the past few years.

INTEGRATED

From smartphones and tablets with their built-in cameras up to desktop monitors and ultraportable laptops, a large number of computing devices now come with integrated streaming capabilities.

After all, the only requirement is a camera. Many devices now have two—one forward-facing and one on the back of the device—and some form of audio capture. The most recent laptops from Apple and HP even have dual microphones, as a way to listen for and potentially eliminate ambient or background noise in a quest to make speech capture that much better.