The Conference for Video Engineers

Mark Watson

Netflix

Next year will see widespread availability of a new creative canvas (High Dynamic Range, Wider Colors), a new royalty-free codec (AV-1) and a new royalty-free image format (AVIF). These new technologies offer great opportunities for creating new immersive experiences, both in content itself and in discovery/user interfaces. What are some of the challenges when we move from "find, click, play" to a rich, seamless and dynamic auto-playing user experience ? How will these new technologies help - and hinder - that evolution?

Yueshi Shen

Twitch

In this talk, we will examine FFmpeg as a live-stream RTMP-to-HLS transcoder and provide some insight on how to operate the tool. Such a solution is simple to deploy but has a few technical issues, such as segment misalignment, unnecessary performance sacrifice, and a lack of flexibility to support our product features. Therefore, Twitch has implemented its own in-house transcoder software stack called TwitchTranscoder, which runs in a custom designed threading model and outputs N variants in a single process.

Tarek Amara

Twitch

S-frame (Switching Frame) was added to the AV1 video encoder to allow shorter segments in ABR, enabling low latency live streaming without the burden of using bit-consuming IDR frames. Segments will be allowed to start with S-frame and video player could switch, decode and play at any segment boundary. This talk will cover specifics of S-frame and show how up to 20% bitrate savings could be achieved on top of AV1 encoder.

Megha Manohara

Netflix

Netflix is available all around the world, and we want all of Netflix's subscribers to stream the best quality videos possible in their bandwidth.

This talk will focus on the research and implementation challenges of Dynamic optimizer - an innovative shot-based encoding tool that raises the perceptual quality of streams at bitrates as low as 250kbps.

Hector Ribera

Marton Ribera Schumann & Chang LLP

Many patents have been asserted to cover various aspects of the HEVC technology. While patents for earlier standards have more or less converged into a primary patent pool, patent holders claiming ownership of various HEVC features seem to be drifting in multiple directions. We will discuss the state of the various patent pools forming around the HEVC standard.

Derek Buitenhuis

Vimeo

Think you know all there is to know about different video formats, codecs, and containers? Your users are here to prove you wrong! Let's take a quick dive the insanity of petabytes of user uploads on a top video website.

Mario Guggenberger

Bitmovin

360-degree videos are becoming mainstream media and promise an immersive experience. Adjusting the video viewport according to the user or device movements is just half the battle. Not only video but also audio is an important factor for an immersive experience. In this talk I will show how ambisonic audio can be handled in JavaScript using the WebAudio API, to enrich 360 video experiences with true "surround" sound.

Gary Katsevman

Brightcove

The w3c has two main caption and subtitle formats that they maintain specs for: WebVTT and TTML. We'll go into a quick history of the two formats, how they look like, what features they provide, and how to use them on the web. We'll end with a verdict on which one is better!

Kieran Kunhya

Open Broadcast Systems

As you walk the halls of major tradeshows such as IBC or NAB, you’ll see that every booth talks about why IP is the future. But they’re not just talking about delivery, they’re talking about using IP for live production of content instead of current technologies such as satellite and SDI. This presentation will explain how these industries are changing and go into some of the engineering challenges of handling tens or hundreds of gigabits of uncompressed video on standard server hardware whilst interoperating with traditional hardware.

Sytse van Geldermalsen

Unified Streaming

With the rising trend of stream personalization, we present a true scalable server-side stitching solution for all major playout and DRM formats.

The architecture is stateless, pull oriented, and also features just-in-time transcoding to fix non-matching content.

The presenter was responsible for the just-in-time transcoding component facing various challenges relating to segment based transcoding with audio and video.

Allison Deal

We'll talk about the standard for live linear advertising signaling and the challenges around adapting these traditional broadcast practices for the OTT world. The discussion will focus on strategies to ensure an accurate and consistent customer experience and techniques for real-time monitoring, identifying, and mitigating inconsistent metadata across hundreds of channels.

Dr. Greg Mirsky

Beamr

This talk will be devoted to the practical aspects of HDR technology for video engineers. Most HDR presentations these days focus on color spaces, EOTF and OETF and tone mapping, but this one's for the practical aspects of HDR deployments. We'll cover the pros and cons of various formats from a practical deployment point of view, the proliferation of various flavors, CPE (Customer Premises Equipment) as of 2017, constraints, workflow aspects, common dos and don’ts, and HDR's impact on codecs.

Matt Ward

Mux

Content delivery is hard and there are many options out there which can feel overwhelming at time. What actually are all of those options and how do you decide what to actually do? This talk will go into the details of how to make CDN decisions and actually improve the things that you once thought were out of your control.

Steven Robertson

YouTube

Tuning ABR starts off with broad strokes, but perfecting it can require a second-by-second understanding of what your algorithm is doing. We've built a system that captures millisecond-level detail on what's going on during a playback (easy), and invested in figuring out how to display that information effectively so we can spot useful improvements (harder).

Scott Kellicker

Wowza

We present our work on low latency streaming: how we achieved 3-second camera-to-screen latency at scale. We'll talk about the player and server technology we developed to stream over WebSockets, and how we are going about scaling it as a worldwide service.

Jamie Sherry

Wowza

We present our work on low latency streaming: how we achieved 3-second camera-to-screen latency at scale. We'll talk about the player and server technology we developed to stream over WebSockets, and how we are going about scaling it as a worldwide service.

Rik Heijdens

JWPlayer

We've implemented a scene detection framework that segments videos into logical story units without the need for a human editor. We achieve this in two steps: 1) we train a deep-learning model to learn a distance measure (i.e. similarity measure) between all pairs of shots by leveraging visual, audio, and textual features extracted from the video; 2) we then cluster contiguous groups of shots into scenes based on the full similarity matrix of shots.

Vittorio Giovara

Vimeo

This talk aims to shed light on colorspaces - what they are, how and why they work, why we should care about handling edge cases properly. Starting with historical design choices, venturing through current standards such as BT.709, and arriving at modern times with High Dynamic Range, the focus will be on practical applications on the web and in broadcast.

Reinhard Grandl

Bitmovin

Since Netflix started using ‘per-title encoding’, optimizing the encoding recipe for the content has become a well-known concept. But what if a similar approach could be applied to every scene on the client side using the adaptation logic? Many low complexity scenes do not need to be streamed at a high bitrate to maintain a consistent perceptual quality for the viewer. These sequences can be replaced by lower bitrate segments, saving bandwidth and reducing costs, if the client is aware of the video context and not only the bitrate of a segment.

Thomas Daede

Mozilla

AV1 Update