Encoding & Transcoding 2018: Part 3
Capella, with 12 employees in San Jose and a small sales and marketing team in Tokyo, was founded in 2010 by alumni of Rhozet, the pioneering video encoding company that was purchased by Harmonic in 2007.
Capella offers its Cambria FTC file-based encoder as well as its Cambria Live encoder, each of which account for about half of the company’s revenue. Both are software based. Capella’s customers include broadcasters and post-production houses as well as content providers, some of which submit to Netflix. Capella also shares customers (e.g., TV stations in Japan) with online video platforms like Brightcove.
After Rhozet's acquisition, Capella’s founder and CEO, Ikuyo Yamada, saw opportunity with Tier 2 and Tier 3 customers that wasn’t being exploited by larger players in the encoding space.
“A lot of big players are still focused on traditional Tier 1 customers like broadcasters,” says Yamada. “But because the technology [and fast internet] is so available, it allows non-broadcasters to do a lot of live streaming.”
“For Tier 2 and Tier 3 customers, the traditional products developed for Tier 1 customers are very difficult to use,” Yamada adds. “My goal is to create a product where if you know how to work a PC then you know how to use our application.”
Capella’s small size allows it to be agile and cost effective, which in turn enables it to meet the needs of less conventional customers. For example, Capella does live streaming from churches in Korea as well as hotel wedding venues in Japan.
Yamada also saw that server-side ad insertion (SSAI) presented monetization opportunities heretofore unrecognized. “Cricket events in India go on for five days and have very, very high viewership,” she notes. “With SSAI, customers can offer live streams and easily make millions of dollars. Our Cambria Live is being used for those type of events.”
Yamada also sees monetization opportunities in many niche events that have not typically been broadcast but can be streamed OTT, from quarterfinal matches at Wimbledon to minor league baseball games in Japan.
Capella’s Cambria Live encoder was used to stream FIFA World Cup events throughout Japan. The company’s Cambria FTC file-based encoder supports up to 8K HEVC (no HEVC for Cambria Live yet), and has been used mostly for events and demonstrations in Japan.
Here again, Yamada senses opportunity. She notes that if she can get her product to be good enough to satisfy the very quality sensitive Japanese market, she is confident she will be successful in the U.S.
Madrid-based Epic Labs was founded in 2015 by four engineers from Akamai and EMC. The company aspires to help media companies transition from a broadcasting world to an IP-based world, and in the process leverage machine learning, advanced encoding algorithms and innovative packaging techniques to create a higher quality of experience for streaming media.
Epic’s core product is its LightFlow Engine, which features per-title, per-scene encoding to achieve low latency, high QoE encodes. Epic’s value proposition is to achieve high-quality per-title encodes, both VOD and live, at an affordable price.
“For live, we apply exactly the same algorithm that we apply for VOD,” explains Guillermo Ayala Benito, Epic Product Manager. “The only difference is that we let the live feed run … sample it in parallel, analyze it, and then after 90 to 120 seconds we come up with an optimal encoding configuration. We then push that configuration into our encoder and you start benefiting from the optimal configuration.”
Epic breaks certain predictable content up into a playlist of titles. For example, a football match might start with talking heads (title 1), then the match itself (title 2) followed by a commercial break (title 3). The match would then continue (title 4), perhaps followed by post-match talking heads (title 5). Epic would apply different encoding profiles to each type of title, with lower bitrates for static talking heads and higher bitrates for high-motion action, like the match.
“Our customers want to get the bitrate down, but they’re very interested in quality,” says Benito. “In many cases, they are IPTV players or broadcast players that are entering OTT for the first time, and they have the problem of the prestige. If you are the BBC and you go over the top, you are still the BBC.”
EuclidIQ is a Boston-based video compression engineering firm with six full-time employees, four of whom are PhD video engineers (they also hire independent contractors). EuclidIQ’s focus has been entirely file-based and runs on pure software in the cloud (AWS).
While many video encoding companies focus on the sexier challenge of UHD, EuclidIQ works with service providers in emerging markets (think India) to deliver the best quality video possible over sub-optimal networks.
“If you try to fit a 5Mbps video into a 1Mbps network, then it’s going to look really bad,” says Nigel Lee, EuclidIQ’s CEO. “We’re able to deliver video in very bandwidth limited situations where they’re still on 2G and 3G and under 1Mbps. You have to figure out a way to present a satisfactory video, not necessarily pristine, in these low bandwidth scenarios.
High-bandwidth broadband has yet to reach many corners of the world, presenting a wealth of opportunity for companies like EuclidIQ. Lee notes, “We are 100 times smaller than some of the other companies out there, but we focus on doing something different from what other people do.”
Delivering acceptable video in these environments presents engineering challenges as great as those faced in the 8K UHD world. EuclidIQ’s team of engineers have developed their own AI algorithms through perceptual quality optimization to achieve the desired results.
“What we’ve done in our company is make sure that we provide good data for the AI algorithms,” says Lee. “That’s what we believe distinguishes our CAE from others. We have better data that we put into it.”
When selling its services out in the field, EuclidIQ often finds itself comparing “apples to apples,” rather than demonstrating stunning encodes through high bandwidth networks. “Most companies [have whatever codec they’ve been using] and then they compare against ours [in real-world, low-bandwidth situations],” says Lee. In many cases, those are competitions that EuclidIQ is winning.
NGCodec is a 6-year-old startup based in Silicon Valley that does one thing - build hardware-accelerated live encoders based on FPGA. NGCodec’s RealityCodec appliances live in the cloud (AWS, Alibaba, Huawei) and are available through a software-as-a-service model. The company’s customers are those who want to build their own workflows using open source software.
NGCodec focuses exclusively on video encoding—no audio, no subtitles, no DRM, no packaging—and delivers AVC, HEVC, VP9 and AV1 encodes with multiple ABR outputs. Gunasekara confidently notes, “We believe that we have the world’s best encoding technology.”
Gunasakara has made a number of decisions that separate NGCodec from the encoding pack. While the great majority of encoding companies now focus on software solutions, NGCodec has cast its lot with relatively new (about three years old) FPGA technology. FPGA offers more flexibility than ASIC hardware, yet delivers higher performance than software-based solutions.
“We’ve gone against the grain,” notes Gunasakara. “Ten years ago, hardware was the model because you couldn’t do it in software. Then five years ago people started moving to software and they thought that that was the right answer. We went in the complete opposite direction six years ago and [ultimately] chose FPGA.”
Gunasakara likes FPGA’s combination of flexibility and performance. “Standards are getting so complex … and if you try to do live video with software you don’t have enough performance … By using hardware acceleration and doing live, we can drastically lower the bitrate.”
While NGCodec is actively involved in HEVC and AV1 R&D (as are many encoding companies today), the company is one of the few that still sees life in VP9, particularly in the exploding eSports space.
“Esports is already a huge part of our business, both here in the west and in China,” says Gunasekara. “It’s a massive phenomenon in the younger demographic… we’ve done quite a lot of optimization in our encoder to make it really good at synthetic or computer-generated graphics.” NGCodec has built a strong reputation in the eSports space with its VP9 encoding, and is poised to announce some major market success in the coming days.
According to Cisco's Visual Networking Index, IP video traffic will make up 82% of all global consumer internet traffic by 2021, up from 73% in 2016. Additionally, by 2020, there will be 26.3 billion networked devices globally, up from 16.3 billion in 2015. With 5G on the horizon, the quality and volume of mobile video will continue to impress.
Over the next year, in addition to this continued growth of online video, we can expect to see reductions in latency of OTT streaming, continued codec efficiency improvements, innovations in CAE, and new applications of AI and machine learning for encoding and other aspects of the video workflow.
As consumers demand a better and more consistent QoE, companies will strive to improve the metrics (PSNR, SSIM, JND, etc.) that they already use to define "quality." But as Nigel Lee of EuclidIQ points out, "What the computer algorithms in the video industry are trying to do is mimic what humans are seeing … In this industry, the human is still the gold standard."
We can also expect some surprises. For example, how many predicted a year ago that one of Azure's largest providers of video content for storage and access would be Axon, a company that this year uploaded 20 petabytes of police department bodycam footage to the cloud?
“The security camera industry is all going to go to the cloud. That’s where the next generation of codecs is really important," notes Kieran Farr, Bitmovin’s vice president of marketing. "You're only accessing this content very rarely, so you want to make sure you're storing this in the most efficient way possible."
To quote Capella’s Yamada, "When there is a lot of change, there is a lot of opportunity, and it doesn’t come so often. We’re in a very interesting time.”
Continuing our look at the major players in the encoding and transcoding space, we look at what ATEME has been up to.
We continue our survey of the encoding & transcoding market with looks at Encoding.com, Bitmovin, Brightcove, Beamr, Synamedia, and Ericsson/MediaKind
Encoding and transcoding are at the heart of every OTT and online video workflow. The first article in this three-part series gives an overview of the technologies and a look at three major players in the space: Harmonic, AWS Elemental, and Telestream.
Now widely available, per-title encoding makes whatever codec publishers are already using more efficient by creating a custom optimized encoding ladder.
An insightful new service called Mux Data makes quality of experience monitoring and analysis easy. This illustrated guide explains how to use it when diagnosing problems big and small.
If you're not already using per-title encoding, it's time. Here's a guide to choosing the tool that's best for you.
A new generation of encoders looks at the context of content to deliver better video playback and higher efficiency. Here's what publishers need to know about CAE.