It's the *last* meetup before Demuxed! It's also probably the last meetup of the year, since Demuxed is in October, and November and December have some minor holidays in them. As usual we'll plan an informal holiday party (hopefully with less bar owner having a conniption over grilled cheese smoke), so keep an eye out for more details there in the next few months.
Given the timing coming off of IBC, we've got a *very* topical talk from Tom Vaughan of x265/Multicoreware/Beamr fame. We'll be live streaming again, so we'll post the details here and on Twitter!
Tom Vaughan - The state of advanced codecs; separating hype from reality.
Tom Vaughan is VP Strategy at Beamr, a leading codec developer. Before joining Beamr in January, Tom was VP Video at Multicoreware, where he started and led the development of x265. Tom will update us on the state of HEVC and other advanced codec projects, while covering both technical and business considerations for adopting these new standards.
If you haven't been to the Demuxed website recently, you should go check it out! We'll start making public announcements next week, but there's a slew of speakers that may have made their way onto the website already. Spoiler alert: there's a ton of really great content.
This month is in the more-than-capable hands of the venerable Nick Chadwick, so if you have any questions ahead of the event, ping him on Twitter (@chadnickbok) or message here on Meetup.
Without further ado, our 2 top notch speakers/talks this month!
Yueshi Shen (Twitch) - VP9 Transcoding for Live eSports Broadcast
To reduce the operational cost, Twitch is building a VP9 live-encoding solution that can deliver broadcast-level video quality with 25+% bitrate savings. Based on our feasibility study, FPGA is the only available solution in the market right now. On the other hand, future high-density solutions can be based on ASIC if there emerge more cost effective encoder SoC products.
Yueshi will explain the thought process of selecting VP9 as Twitch's next-gen video codec format, particularly will elaborate 1) legal risk, 2) decoder support, 3) compression gain.
Zhi Li (Netflix) - VMAF: the Journey Continues
VMAF started as a research project to address the need to automate quality assessment of Netflix's vast video catalog. In June 2016, we open sourced VMAF on Github, and since then it has gone through some substantial updates. In this talk, I will focus on its latest improvements and enrichment, including speed optimization, more accurate models to predict mobile and 4KTV viewing conditions, and adding a confidence interval to quantify the level of confidence in the prediction. Lastly, I will give a sneak peek on what is coming up next in VMAF.
# Demuxed Ramblings
It's July and the talks are in! We've officially hit *100* submissions this year! If only meetup allowed emoji, because you'd see a few party parrots right now. The selection process is starting any day now, so hopefully we'll have something to show for it in the next week or two!
Christian Feldmann (Bitmovin) - "Scalable Video Coding in HEVC & AV1"
While it has never played a big role in practical applications, scalable video coding has been around since the times of mpeg2! 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, we 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.
In case you haven't noticed, some people are playing soccer *without* the US, but the world seems to have not noticed and are still watching. I know, I know, it's really wild, but in the meantime there's a _lot_ of international football being delivered over the web right now.
Speaking of international smash hits, Demuxed talk submissions close at the end of June! Yes, that means you have just a few more weeks to cram the submissions full of your incredible talk abstracts so we can assemble the greatest Demuxed lineup of 2018.
We're looking for speakers for the next few months! If you, or someone you know, would be a great SF Video speaker, please nominate them (or yourself): https://airtable.com/shrVeg3DuzUhPiKd5
Just to note, you can also consider that global. If you know someone in NYC or Seattle, for example, you can still have them fill out the form and their name will get passed to the relevant organizer :)
Hadar Weiss (Peer5) - Tales from delivering the FIFA World Cup Peer-to-Peer
Haluk Ucar (IDT) "Running live video with FFmpeg"
A discussion on specific considerations when using FFmpeg for live encoding at scale and high, consistent quality
I've already made the "May showers" joke in the last 5 years for a meetup description, so...I'm running out of month-specific material here, y'all.
[Demuxed] Monthly reminder that the call for proposals for Demuxed is *open* and you should go submit all the crazy ideas you've got. If you do it earlier, someone will even give you feedback to make it better for the selection committee!
[Meetup Stuff] This month's meetup is courtesy of someone that grabbed me after a meetup and said they had cool stuff to present! My sneaking suspicion is that a lot of you have ideas you're not bringing up to me, so to make it more structured, I made a form! https://airtable.com/shrVeg3DuzUhPiKd5 Give it out to whoever you'd like, and if you know someone that lives in another city, still put them in and I'll share them with the organizer that's nearby.
Andrey Pozdnyakov (Elecard) - A simple way to estimate internal video camera noise - Gaussian and salt-and-pepper noise.
The method described could be used to compare cameras, dependence of noise level on illumination, camera temperature, or even internal noise with compression artifacts of lossy compression. We'll conclude by talking about what is the level of compression that doesn't affect image quality more than its own internal noise. In fact, some level of compression could bring even *better* image quality.
Psst Demuxed.com now has shiny links to both submit your proposal _and_ buy tickets for Demuxed! The early bird tickets are going fast, so if you're planning on coming, don't dilly dally. Reminder, if your talk gets accepted and you already have a ticket you'll get a refund 🙌
Ryan Damm (Visby) - Holographic update: Light Fields and the Future of Video
The future is holographic, from VR/AR to next-generation holographic TVs. But where are we today? Let's take a deep dive into next-generation video standards, as well as the current state of the technology and the major players.
Potentially one more familiar face as well :)
As Kieran said in the comments, it's a testament to this community that we can have a waiting list on a meetup with the incredibly insightful description of "TBD"...
The even better news is, it's definitely already determined! This one's been in the works for months and it's finally happening.
Can Bal (Twitter): Introducing Vireo
Colleen Henry (Facebook) and Alex Lindsay (PixelCorps): High End Live Streaming End to End: How to build your own live streaming infrastructure using free software, AWS, and off the shelf hardware.
It's February and this is, at least in my memory, the wildest month for the global video meetup community. I could be off (timezones are hard, you know it), but I think there's been a meetup in at least 6 different time zones!
If you're on the Demuxed mailing list you saw this lineup a few weeks ago...could have sworn I updated it here too, but obviously not. Fear not, though, it's a good one!
Anton Kast - Video on the move: Mobile phones and video
A history of player API's on Android and iOS (including HTML5 video and mobile Flash). The joys of resource limits related to hardware decoding and metered bandwidth, and stream caching strategies.
Alex Castillo - Building a Mind-Controlled HTML5 Video
The human body is a ocean of electricity. Billions of neurons are constantly engaging as our thoughts go from one state to the next. Come see a live demo on how we can capture brainwaves in real-time, get cognitive metrics from our brains, send the data to the browser, and map it to the playback of an video element. In other words, let’s build a mind-controlled HTML5 video!
It's 2018 and we're back to our regularly scheduled programming! If you missed the holiday party, I'm very sorry that you didn't get to enjoy incredible grilled cheese and the delightful company of your fellow SF Video-ers.
Yes, now that it's officially 2018, that means you get to hear me talk about Demuxed 2018 until...basically November. We'll be taking over Bespoke at the Westfield on October 17th and 18th (yes, you're reading that as 2 days). Two days in a more expensive venue means we need amazing companies to be our partners in crime (jk there's no crime, we're not doing an ICO or anything). If you think you (or your company, rather) might be interested in sponsoring, shoot us an email at [masked] .
Nick Chadwick (Mux (https://mux.com)) - Watching Twitch live stream scores with Tensorflow
At the end of last year Mixer (https://mixer.com/), a Twitch competitor, introduced a new feature: Hypezone. Hypezone is a channel that always broadcasts the Player Unknown's Battlegrounds stream closest to winning, and has consistently been one of the most watched channels on Mixer. Inspired by the technical challenge, I built a simple version of Hypezone (Hypebot3000) that does the same thing for Twitch streams.
In this talk, I'll outline how I approached solving this problem, from efficiently watching twitch streams and struggling with OCR, to finally creating a working version using Tensorflow. I'll go over the basics of screen capturing live streams, pre-processing them, and implementing a simple Tensorflow model to extract the current game score.
It's time for the annual SF Video holiday party!
This year we're going to have an open bar and food at everyone's favorite (shh, you know you love it) after-meetup dive bar: Club 93. This is even more informal than most meetups, so just make sure to bring yourself and, if you're up to the task, your favorite ugly sweater.
There may, or may not, be a game of Rock Band happening on the large cluster of TVs in the back of the bar. Yes, we're going to buy all the best (worst) Christmas songs we can find.
See everyone there!