Google Duo rolls out realtime AV1
blog.google
blog.google
Though, AV1 is way, way better than h264 when bitrates are super low... I don't know. I wish I had more details about the implementation.
I will try to be more specific. Youtube uses x264 to encode videos at a certain bitrate. Were you to encode the same video with libaom (AV1) at the same bitrate, the quality difference would probably be noticeable to a casual viewer. If you were to cut the bitrate in half and run the same comparison, the quality difference would be extremely noticeable to a casual viewer.
EDIT: Err, I'm stupid it says 30kbps right there. Though that definitely looks closer to 100 to me (unless framerate is lower)
For making encodes for personal use, HEVC/x265 is the clear choice today, and it probably will be for a long time.
If you want to use AV1/libaom and see significant efficiency gains over HEVC/x265, you will have to allocate a lot more encoding time. There's no way around that. You will also not be able to make use of multithreading to a significant degree, for libaom at least (I have not seriously tested other encoders). There are third-party programs in active development that perform chunked AV1 encoding for the purpose of utilizing all your cores.
aom --cpu-used=5 is currently strictly better than x265 --preset veryslow, putting aside the threading issue. Beats it in encoding time and efficiency. But it's an unfair comparison.
More info: https://www.videoproc.com/media-converter/av1-vs-hevc.htm
It also probably isn't 30 fps.
Also find it funny they use a 2.5MB GIF to show a demo for ultra video compression.
https://storage.googleapis.com/gweb-uniblog-publish-prod/ori...
GIF allows animation and lossless compression.
Cisco also has a real time AV1 encoder. They say it's all about using the right AV1 video coding tools in the right way for the live video use case:
https://blogs.cisco.com/collaboration/cisco-leap-frogs-h-264...
The presentation:
Calculations here, someone please tell me I'm wrong:
The caption quotes "30kbps", but that has to be a typo - I think they mean 30KBps (240kbps).
58.7MB * 1024 = 60,018KB 60,018KB / 30KB/s = 2000 seconds 2000s / 60 = 33 minutes
If it's clever and it doesn't work, it's surely clever ... but it doesn't work
Chat + Forums + Fileshare + Video chat + Voice chat + OneDrive integration + Sharepoint integration + Kanban + goodness knows what else
Before the lock-down we all turned our nose up at it as "yet another lame chat app", now it's dominating everything, and it works on MacOS, Windows, iPhone, and Android -- so there's no excuse for anybody to not have it on their phone as well.
There's nothing else that's even close for us and just on my team we have about 60 people on it, and across the company over a thousand. It's almost 100% replaced all the other things that do the things in the list above with the exception of Jira and a few holdouts still using Webex (so there's a dial-in for people outside our org who don't have teams, but we've even been using for meetings across different organizations that have teams and it works perfectly fine).
Duo, Meet, Hangouts all seem really spare and kind of old fashioned now.
Sharing code requires 4 clicks and there is no keyboard equivalent way to get to it.
And the list goes on.
The sad truth is that if teams didn't come for free with office 365 it wouldn't be able to stand on its own.
You can create private channels within teams. When you go to "Add Channels", there's a Privacy option that's set to Standard by default, but you can toggle it to Private and invite only specific team members to join.
otherwise i kinda agree except it's years ahead of lync, so for microsoft shops it's still a net win. my personal pet peeve is lack of a global mute keyboard shortcut (win+f4 for lync).
I can't call up my Mom on Teams. I can with Duo.
I've never used Teams, but it seems like an awful lot for one app to handle. But the Google way of splitting it up across the logical apps seems to work well, since everything's integrated anyways.
At the end of the day, does it really matter if you're using different icons in your home screen or different icons in a menu?
When Google releases a new product, it's like the almost expect you to use it in isolation, but with your same account. And they love releasing new products. Google Calendar, Tasks, Keep, etc.
Microsoft Teams brings together Office 365, Exchange, Share point, OneDrive, and then adds collab features on top for comms and organisation.
You get your full existing Outlook Calendar appearing in teams, you get all your Outlook contacts, you can link to any doc in your OneD, shared edit, you get group chats and channels, you can create "Teams" (where the name of the app comes from) made up of different people in your Exchange org.
For anybody using Microsoft AD/Exchange within their org, Teams is a natural choice. You don't need a new account, everybody else in the org is immediately searchable and contactable, there's no data import needed. It just works. And "it just works" has been the central driving feature of anything that has taken off during this lockdown.
Now that we've all been forced to use it and come to terms with it, the deep integration is extremely powerful. In fact the couple areas within it where the integration is still a little thin stand out because the workflows around those end up being a bit fussy.
It can be difficult to explain to somebody who doesn't use it, but it's not something Google has sorted out yet, not even close. It's like saying "I have all the ingredients for a cake in my pantry, so they're all together, it's basically the same as having them together in cake form."
Google has all the pieces to do something even better than this, but their corporate culture is like a hammer shattering all the pieces into different bits with only the most superficial connections between them. They may be allergic to even try given what happened with Wave, but even a half-decent copy using the building blocks they have would be pretty great.
^ Candidate for all-time best illustration of the phrase "to damn with faint praise".
We used slack before and teams definitely feels like a step up.
Unless you are one of those people (like me) who don't want to put work related things on your phone. I like to have a separation between work and personal life. Thus why I don't put work email (or other things) on my phone.
If they got rid of the excessive whitespace at each side of the chat and used libnotify like civilized folks I'd be super happy.
I wouldn't hold my breath. They use the same custom notification on Mac os and even on Windows.
Oh, and on Linux, have you tried using it with a tiling wm ? The notification gets its own regular window. The first time it happened it was quite jarring to have it fill half my 32" monitor. Even since that happened, I've went back to using it through Firefox, which honors the system notification setup.
I can cross over to their full videoconference with an invite, share screens, put notes in chat, whole 9 yards. For a moment, The company is just a logical division of the whole Teams infrastructure.
That's powerful.
I've never experienced anything quite like it even if on paper it sounds like half a dozen other conference systems. It blows anything else I've used out of the water.
In IT company where Mac and Linux dominates we are already accustomed to other software and don't see any benefit in changing everything to use One Drive.
What google needs to do is to stop fracturing communication modes across every single conceivable strata they can identify.
Pricing: https://www.microsoft.com/en-us/microsoft-365/microsoft-team...
If I could junk webex entirely I'd love it.
If Google couldn't care less to bring it to my attention, I couldn't care less to try it.
And now it's suddently important to them?
Do the right thing at the right time. Invest and believe in your product becuase it's great.
Don't try and jump on the bandwagon now that you've just suddenly reversed your decision to sunset the product because of what's going on.
/rant
Meet == Zoom. It's video conferences for businesses.
Duo == FaceTime. It's video calling, with integrations into the Phone app on Android.
Then they decided that wasn't the way they wanted things and have been slowly killing it off for 5ish years now
There is a timeline, at least for entreprise customers.
> Classic Hangouts will be retired for all G Suite customers starting in October 2019
Chat within Gmail is still hangouts. Hangouts.google.com still works. Users still use the hangouts apps.
On the personal side, if you use Google Fi, SMS integration is still only available via Hangouts. This is a somewhat unique feature of Fi and Google Voice. You can send/receive SMS within Hangouts via browser or app all the same. It's not like the Messages app for web (which relies on your phone being online).
They have a newer timeline at https://support.google.com/a/answer/9197126?hl=en&ref_topic=...
I think it's still using some proprietary transport which is UDP over TCP over UDP or something...
TBF both Meet and Duo have been awesome, Hangouts has withered on the vine. It could have been a unified solution but... Google.
My understanding is Google has said externally that they plan on killing the classic Hangouts product. So that means Google chat and Google Meet would be available in consumer space for people with @gmail.com addresses. but right now the focus of those two products is on gSuite.
I'd also definitely agree with what other people are saying, duo is the FaceTime competitor. It has direct integration into the Google dialer in the Google SMS app. It's also rather brain dead simple to use.
This is weird to hear, to me hangouts was always the new thing that ousted gchat.
> google chat [...] would be available in consumer space
I can't imagine they're bringing back the old old thing, are they?
The Wikipedia disambiguation page really clears up the mess that is Google's branding: https://en.wikipedia.org/wiki/Google_Chat
https://www.theverge.com/2020/4/9/21215588/google-chat-hango...
It is effectively a slack or teams competitor.
Classic Hangouts is/was a lot of things. Sadly it was too many things. The new Chat has very similar Gmail integration, so if they ever get it out to consumer, it hopefully won't be a big change for people.
Specifically, where do the private keys reside?
Desktop -- Meet/Chat converging and getting a unified UI for business/desktop users.
I think it is taking time, but it's finally happening and Hangouts will officially be shut down when the merging happens.
The integration with Meet consists of a single "invite to video call" button that creates a new video call and inserts the link into your message. There's no option to stop people from inviting you to new calls/set your Google chat status as busy when accepting a call.
Meet itself is also frustrating to use for anything more than a 2-3 person meeting; it doesn't display more than 4 video feeds at once on your screen. It's also unable to save settings between video calls (if you don't want to have incoming and outgoing 720p streams all the time you have to change the settings at the start of every call). Zoom and even Jitsi are both miles ahead of it.
There is no reason to use either product besides the fact that they come with enterprise Gmail accounts. They both still feel like prototypes or minimum viable products (especially Chat).
A mobile only consumer proposition is useless. Sounds like Google's internal strategy here is still a disaster.
The "knock" feature is kinda novel and weird (in a good way). When you call someone, they can see your video feed before they answer the call.
Whereas solutions like Meet and Zoom are designed primarily (both UX and architecture) around multi-party, supporting 100's (250 for Meet, 500 for Zoom).
But under the hood, limiting the number of users allows some important technical advantages. Duo claims (in its FAQ) to be end-to-end encrypted. I believe Whatapp claims the same. And it wouldn't surprise me, knowing Apple, if their video chats on Facetime did too. This is possible only with a reasonable limit on the number of participants because of the need to negotiate a bandwidth for each stream that works for all viewers, and each stream needs to be independent, not multiplexed into a single live view.
The proper use of large-scale meeting products is in corporate or paid environments where you have some kind of guarantee that your chats aren't being recorded or sold. (At least unless a three letter agency is involved.) Tools that guarantee end-to-end encryption should be used in small group chats.
Your point about UX is a strong one too. There are a affordances that work better for smaller chats, others for larger ones, and it's helpful to be able to specialize.
In banking/financial services, we want guarantees that all chats are being recorded. In lots of other regulated and government fields too.
Because beyond UX, there are a ton of technical tradeoffs involved.
In a 2-person call you want to avoid a server if possible, because it adds latency -- so that's a good first reason. Zoom and Meet would never dream of not using a server.
The optimal architecture of a "small group call" is totally different from a "large group call".
I agree though, this feature seems like a solution in search of a problem. If I'm answering the call it seems more natural to activate both cameras at the same time, and for calls I'm not going to answer the video feed just seems unnecessarily distracting. I'm sure that I have never wished for a feature like this.
Seems like the sort of thing that perverts would take advantage of.
Presumably you would know who you're video calling anyhow prior to calling them, so there's not really a meaningful privacy violation...
The only case I can think of is you calling me to expose yourself or something - to that, I'd say that a privacy option where you can make it so you have to add someone before they can video call you would suffice.
One of the problems with an opt-out setting is that it can become a viable attack vector whenever the application settings reset: "OMG New phone for my 14th birthday" will look like an invitation to some segment of the population.
http://support.google.com/duo/answer/6376115
Google knows about dick pics.
Although to be fair, me confusing Duo Security and Duolingo is because I use both and both have nearly the same green for logos.
I’m definitely not holding my breath though.
This is very worrying because hangouts is the only thing I use. All of my contacts use hangouts except for a few SMS, and I heard about Slack.
And this is a mostly current or ex Google crew...
Basically compared to Skype, Facetime, Whatsapp, Facebook, Signal, Telegram, etc. Duo is an also-ran that most users of the before mentioned competitors have probably never even heard off. I understand they managed to get some traction in some secondary markets that are mostly not the EU or North America. I actually have conversations with people on all of these, except Duo. I've never had a single person reach out to me with the request to do anything whatsoever on Duo. Not a thing. I've never seen it in the wild. I don't know a single person using it.
I just came out of a meets/hangouts meeting. I'm sure the rebranding is very significant to Google internally while they deal with Conway's law and internal teams continuing to try to "own" this space. But from the outside it looks like more of the confused strategy around Google Voice/Hangouts/Docs/Buzz/Wave/etc. where there's always an orphaned thingy that is actually useful that vaguely enables people to actually talk and chat to each other.
Meanwhile, between Slack, Microsoft Teams, Zoom, Google seems to be lacking a good strategy for competing effectively with those. I find it very telling that startups that are Google customers still opt to use things like Slack and Zoom. There's something missing there; a vacuum that Google has been dancing around for years without actually managing to compete. Literally every team I've been on in the last five years had Google docs + gmail as well as Slack.
Glad to see they are still making improvements, but with Google you know that products are simply abandoned from time to time so can't get too attached to it really.
I've been blown away. I've never given it access to my phonebook, so I just dial numbers by number, but the video quality is stunning. The latency is so low as to be almost jarring compared to other video services, and usage is simply effortless. Duo Just Works™. I haven't looked into how or why it's so good, but I'd love to learn more....
....and hopefully find an independent system that works the same way so I can keep using it when Google inevitably shitcans this a few months down the road.
Maybe they can teach the Waze team to do this since they're both part of Google. Relevant because in my driveway my phone will still be on my wifi, and when I drive away, Waze becomes unusable for a few minutes.
On one hand, we expect them to innovate and disrupt with new experimental products - then we get annoyed when they discontinue the ones that don't take off. ...and now we are criticizing them for not consolidating overlapping apps.
Do we ask startups not to launch products they think will compete with existing messaging apps? In reality Google operates (somewhat) like a collection of startups.
It's a messy marketplace for apps. Is it really sooo bad that giant Google has a couple different apps for messaging? We should also acknowledge that they are consolidating them (at least Hangouts and Meet).
Maybe they could do a better job, but our negative reactions to them are out-sized.
Google simply has bad product management propped up by the huge legacy cash pile they have from advertising.
Android dialer should have Start Voice Call, Start Video Call, and Start Group Call, where only the first one uses ye olde voice network
Basically, Duo suffers from being only video chat. I don't know why Google can't make a WhatsApp competitor that's good. I think the problem is it's never had a CEO that really cared about this product category so it flails around, dominated by professional product managers who don't have any burning innovative idea (unlike when WhatsApp was new and the big idea was to use phone numbers instead of user names).
[0] https://play.google.com/store/apps/details?id=com.google.and...
Current availability: Chat features in your Messages app are available in France, Mexico, South Africa, Spain, the UK and the US, and across multiple operators globally. We are working on bringing chat features broadly to more regions, operators and other messaging apps (e.g. Samsung Messages).
https://arstechnica.com/gadgets/2018/12/google-promises-to-s...
Apart from GMail, the only Google communication app that actually made a difference and people liked was Hangouts. For unknown reason, Google would really love to kill it. It officially does not even exist anymore, but I and many others still use it daily.
Funny how when you have nearly infinite money you can do absolutely mad stuff and no one just cares.
https://arstechnica.com/gadgets/2019/04/googles-constant-pro...
Unlike chat apps, there's not that many people I want to video call.
At some point clear marketing becomes a user facing feature.
Why would she care? It's a separate app. Just fire it up and start a videocall with someone. With Duo I can call friends on Apple devices, and the quality of the call is excellent. I use it to call my mom (in another city) any time I want to talk to her. She's a technophobe, but even she can answer the call and start chatting with me.
There’s frustration on a lot of aspects, one of them for instance being that both Hangouts and Meets are integrated to Google Calendar, so if you use both a private and corporate gmail, you’ll get a link to one or the other depending on your meetings. Sometimes from the same calendar screen. Some people won’t care, but there’s enough cases where it will be a PITA to remember which one you’re using or will be using, just to get back to the right window for instance.
But really the main frustration is you can’t give someone a Duo address. Or a Google Meet address. You have to give them your email, and then negociate what service they can/are comfortable using. And ghey have to remember what was negociated.
My understanding is you will be able to receive video calls inside android/messages even if duo is not installed, but initiate calls only if duo is installed.
(As it isn't shipped on Samsung devices.)
Don't know what Google does differently but boy this thing still works with really shitty slow internet. Glad to see AV1 usage here now. HD video chat with an analogue modem connection ;)
Google starting to use AV1 in their products in earnest should either trigger the patent suits from the aforementioned "at least one company", or serve as an additional argument in favor of any company that starts using it now.
Lots of good bits came from FOSS codecs but Daala and friends have never competed with the commercial stuff that have 100s of decades of paid engineering time in development.
Also it’s pretty impossible to do anything in the codec space without tripping over a patent.
That said the very best implementation of the patented and un-free H.264 codec was built as open source (x264) but you still technically need to license all the h.264 patents to use x264 legally in many scenarios or jurisdictions.
If you read some of Xiph's development blogs (with work on new audio or video coders such as aforementioned AV1), a lot of the process is carefully going around the "inventions" that the bug bucks have declared as their property at the patent office.
Even the current state-of-the-art audio codec Opus gets a lot of its efficiency from proprietary algorithms from Skype that were Open-sourced and parents made royalty-free.
Video codecs are bleeding-edge comp-sci and expensive to develop (lots of highly paid engineers, test labs, and test subjects (people) required).
As for video research being expensive, you're right, of course. We're probably seeing much increased funding in FOSS codecs now compared to all previous decades.
In Software, with 720p video (or more), the battery of the phone would be empty in no time.
Still though, battery concerns aside, massive kudos to Google here for helping to advance a technology that is so important to OSS.
It's exciting to see more mainstream adoption of AV1 which will surely drive HW acceleration forward, but it's disappointing that Apple hasn't been more proactive here.
Video encoding is a modeling and search problem. Codecs try lots of different ways to represent the input data and do their best. You can evaluate the costs and benefits of each piece of the encoder to decide which parts to have on or off.
Obviously a speed-tuned AV1 compressor won't achieve maximum efficiency, but it might still perform better than an H264 or H265 compressor since it has more efficient tools available to it.
On large PCs it's been possible for a while, through SVT-AV1. On phones I think less so right now, if you're talking about using AV1 to get noticeably better rates than VP9.
That said, it's possible that they have a reduced set of tools/passess that can run in realtime on a phone, and it looks like the rate is extremely low, which also improves encoding complexity. Even software VP9 encoding has been rare on phones though up to this point, so I'm not sure exactly what is meant by this change.
Yes, otherwise what would be the point of using AV1 in the first place. Many (most?) Android phones already have hardware VP9 and HEVC encoders. iPhones only support HEVC, so you know what the choice has to be to maximize the compatibility. HEVC and VP9 offer pretty much identical subjective visual quality for a given bitrate for the purpose of video calling - I tested a lot.
But anyway, even if you were able to run a realtime software-only AV1 encoder on a phone, there's still the concern of heat and battery life. Mobile devices usually aren't designed to run at 100% CPU load for any kind of extended period of time. Hardware codecs are much more power-efficient.
A codec is more like a toolbox or instruction set. It gives you a bunch of options you can use in order to compress a frame.
Unless AV1 removed some of the tools from the earlier codecs it drew from (e.g. VP9), it's entirely possible to write an encoder with similar performance characteristics as the earlier codecs.
What I'm getting at here is that AV1 isn't inherently computationally complex. Its reputation for being computationally complex comes from people looking at the reference encoder early on during development and treating that as the final product.
The reality is that it shouldn't be too far from HEVC in terms of computational complexity.
As far as I can tell, Google for example tries to keep its custom hardware (as in custom relative to the rest of the infrastructure) very minimal.
But if Zoom or any other actor can for example implement some scalable ASIC or other dedicated form of hardware to transcode and do networking (perhaps even some encryption primitives), I could see how that would be easily done.
Then again, I'd say Google and company have enough resources to do so, just not the incentive. I don't know the details about the finances at Zoom, but I'm pretty sure it costs A LOT to handle thousands of 1000 people conferences. If they're getting paid, and they're just getting started, maybe it's worth the cost. At Google et. al., perhaps is just a matter of cost vs. revenue. But they do have MORE than enough of brains, software and hardware to do so.
I've never understood why Facetime doesn't allow this.
Though if the goal is to allow for extremely low bitrate use it might not chew up too much of the processor and battery.
1) end to end encrypted, and
2) we’re constantly making optimizations
3) uses AI to reduce audio interruptions,
4) and Excitingly! Coming Soon! we’re rolling out a new video codec
5) that works with very low bandwidth connections
I dunno I use FaceTime, it “just works”
<sarcasm> I just tried to make a FaceTime call from my Google Pixel phone and it didn't work :( </sarcasm>
Google's latency isn't stellar either, mind you. For all their issues, I think the best mainstream solution for latency right now is Zoom...
Disc: Googler.
And he would usually his story with: -And I've been an Apple-fanboy ever since.
I miss that dude.
Google doesn't seem to realize it's not easy getting your friends who aren't techheads to move from one platform to another, so in order to switch I have to know this isn't going to die soon. Untill then I'll stick with hangouts/skype. Why couldn't they just keep the hangouts branding? I have no idea, its kind stupid.
(Also yes, I know hangouts is going to die. But it hasn't yet and that's what all my friends already know about.)
"WaveNetEQ is already available in all Duo calls on Pixel 4 phones and is now being rolled out to additional models."
I can reach more non-technical people on Signal than on Duo.
Anyways I welcome the competition.
I've noticed video quality on it seems to be really poor recently and even just freezes for people who aren't the current speaker.
I don't have the same issue with Slack, Teams isn't great but isn't as bad as Meet.
But Duo for more than 2 people is currently Phone only. Also, you can't schedule a call, which could be annoying for meetings. For personal calls I prefer Duo.
I don't have any contacts with it, I guess no one has faith it'll stick around like all other Google chat apps.
Edit: I think it is a perfectly valid Question. I dont see why the Downvote. Not to mention it is against the guidelines.
What is the threat model that starts with "we claim to e2e encrypt traffic but don't so that..."?
Storing all the video isn't feasible, not is analyzing it in real time, so the only sensible threat model is targeted attacks, but the companies have no reason to do that. Governments do, but this makes the companies resistant to governments.
Analyzing the audio stream of a video call is possible in real time, in fact Google meet already does it to offer closed captioning in real time (the only good feature meet has compared to other alternatives)
Is it the call metadata (who/where/when) that valuable?
Or is it just at loss to prevent future competition?
Even system apps can be updated through the Playstore though.