Mux’s Justin Sanford explains the difference between Quality of Service and Quality of Experience; the latter being about the entire viewer experience. Justin looks at ‘Startup time’ showing that it’s a combination of an number of factors which can include loading a web page showing the dependence of your player on the whole ecosystem.
Justin discusses rebuffering and what ‘quality’ is when we talk about streaming. Quality is a combination of encoding quality, resolution but also whether the playback judders.
“Not every optimisation is a tradeoff, however startup time vs. rebuffering is a canonical tradeoff.”
Finally we look at ways of dealing with this, including gathering analytics, standards for measuring quality of experience, and understanding the types of issues your viewers care most about.
The title says it all! Alex Converse speaks here at the San Fancisco Video Tech meet up while he was working at Google discussing the ins and outs of AAC – and since he implemented an AAC decoder himself, he should know a thing or two about it.
Sure enough, Alex delivers by talking about the different version of AAA that have been around since MPEG 2 AAC through to the high efficiency AACs we have seen more recently.
Walking through the AAC Encoder block diagram we look at each of the different sections from sampling, MDCT (a type of Fourier transform) to psychoacoustic processing, stereo processing and more.
We then start to look at the syntax for the way the streams are structured which brings us in to understanding the AAC channel modes, and the enhanced mechanisms for encoding and processing used by the later versions of AAC including HE-AAC V2.
Alex finished with quick look at low delay codecs and a Q&A.
A great, detailed, overview of AAC. Ideal for developers and those who need to fully understand audio.
With the advent of digital video, the people in the middle of the broadcast chain have little do to with colour for the most part. Yet those in post production, acquisition and decoding/display are finding it life more and more difficult as we continue to expand colour gamut and deliver on new displays.
Google’s Steven Robertson takes us comprehensively though the challenges of colour from the fundamentals of sight to the intricacies of dealing with REC 601, 709, BT 2020, HDR, YUV transforms and all the mistakes people make in between.
An approachable talk which gives a great overview, raises good points and goes into detail where necessary.
An interesting point of view is that colour subsampling should die. After all, we’re now at a point where we could feed an encoded with 4:4:4 video and get it to compress the colour channels more than the luminance channel. Steven says that this would generate more accurate colour than by stripping it of a fixed amount of data like 4:2:2 subsampling does.
Given at Brightcove HQ as part of the San Francisco Video Tech meet-ups.
MUX is a very pro-active company pushing forward streaming technology. At NAB 2019 they have announced Audience Adaptive Encoding which is offers encodes tailored to both your content but also the typical bitrate of your viewing demographic. Underpinning this technology is machine learning and their Per-title encoding technology which was released last year.
This talk with Nick Chadwick looks at what per-title encoding is, how you can work out which resolutions and bitrates to encode at and how to deliver this as a useful product.
Nick takes some time to explain MUX’s ‘convex hulls’ which give a shape to the content’s performance at different bitrates and helps visualise the optimum encoding parameters the content. Moreover we see that using this technique, we see some surprising circumstances when it makes sense to start at high resolutions, even for low bitrates.
Looking then at how to actually work out on a title-by-title basis, Nick explains the pros and cons of the different approaches going on to explain how MUX used machine learning to generate the model they created to make this work.
Finishing off with an extensive Q&A, this talk is a great overview on how to pick great encoding parameters, manually or otherwise.
San Francisco Video Tech welcomes Haluk Ucar talking about live video streaming. How do you encode multiple resolutions/bitrates efficiently on CPUs and maximise the amount of channels? Is there value in managing multiple encodes centrally? How can we manage the balance between CPU use and VQ?
Haluk discusses a toolset for Adaptive Decisions and looks at Adaptive Segment Decisions. Here he discusses the relationship between IDR frames and frequent Scene Changes.
Haluk covers a lot and finishes with a Q&A. So if you have an interest in Live Streaming, then Watch Now!
VP9 is a well-known codec, but it hasn’t seen many high-profile, live deployments which makes Twitch’s move to deliver their platform using VP9 in preference over AVC all the more interesting.
Here, Yueshi Shen from Twitch, explains the rationale for VP9 by explaining the scale of Twitch and looking at their AVC bitrate demands. He explains the patent issues with HEVC and VP9 then looks at decoder support across devices and platforms. Importantly, encoder implementation is examined leading to Twitch’s choice of FPGA to provide live encoding.
Yueshi then looks at the potential of AV1 to Switch_Frame to provide low-latency broadcast at scale.
There are a lot of codecs both new and old that are in use or vying to be the next big thing. Tom Vaughan helps us see what they really can achieve and where each one is useful.
Recorded at San Francisco Video Tech Meetup in September, this video starts with a look at a the ‘hype cycle’. Tom places each codec, from MPEG 2 to VVC on the curve before looking at what the barriers to adoption are.
Tom then looks at HEVC discussing which devices can receive it, which can create it, the streaming services which support it and where adoption is likely to be. Finally, HEVC discussion is complete without a look at the HEVC patent landscape Venn diagram.
The focus then shifts to the Alliance for Open Media and their AV1 codec, its patent status and technical progress to date. He then discusses the performance of AV1, HEVC and Beamr against each other.
Almost brand new out of the starting blocks is VVC from MPEG and the Media Coding Industry Forum (MC-IF). Tom explains the aims of the forum and the VVC codec they are creating before taking questions from the floor.
While it has never played a big role in practical applications, scalable video coding has been around since the times of MPEG 2, and might actually have some advantages over the multi-rate transmission often applied today. The purpose of scalable coding is to efficiently compress multiple different versions of the same video in one “scalable” bitstream. Actually this sounds like the perfect solution for VOD and streaming applications, but unfortunately it has some downsides and few vendors ever used it. In this talk, Chrstian will review the basic idea of scalable coding, how it is enabled in modern coding standards and the pros and cons of implementing the technology in streaming applications.
VMAF is a video quality metric created by Netflix which allows computers to indicate what quality a video is. This is an important part of evaluating how good your encoder or streaming service is so it’s no surprise that Netflix has invested years of research into this. Other metrics such as PSNR and MS-SSIM all have their problems – and let’s accept that no metric is perfect – but what the industry has long grappled with is that a video that has a strong fidelity to the source doesn’t necessarily look better than one that less-faithfully replicates the source.
Imagine you had a video of an overcast day and one encoder rendered the video a bit brighter and a bit more blue. Well, for that clip, people watching might prefer that encoder even though the video is quite different from the source. The same is true of noisy pictures where replicating the noise isn’t always the best idea as some people, for some content, would prefer the cleaner look even though some details may have been lost.
As such, metrics have evolved from PSNR which is much more about fidelity to metrics which try harder to model what ‘looks good’ and VMAF is an example of that.
Zhi Li explains the history of VMAF and explains some of the new features which were released in August 2018, when this talk was given, which gives an insight into the way VMAF works. Plus, there’s a look ahead at new features on the road map. This talk was given at a SF Video Technology meet up.