VP9 SVC poc roadmap / Slack


#1

I saw that a new poc was added to repo, vp9 svc and I can’t be more excited and interested.
Can you please give us some more infos?
I saw that you are achieving good result in testing in the chromium monorail.
Is there any way we can use it / test it?
I just cloned and compiled, I can see VP9 packets flowing in my console but nothig appears if I subscribe.
Is there any roadmap? can you give us some little info? anything that can help us organize our work, and obviously, contribute.
I also saw the slideshow handler adaption to vp9 svc. Is a future-needed feature or is usable right now?

I take this opportunity to also ask you another little question: in an old PR, you said me you prefer to not include webrtc lib. Maybe it’s a stupid question, I consider myself a fast learning person but I only have 3 months experience with c++ so maybe I don’t see the point. why do you prefer rewrite everything?

Maybe I’m expanding too much, but what about open an IRC / Discord / Slack #Licode?
I think it can be useful to keep more in touch with the community.

I think we all need more involvement to help you guys! I really want to help but often I doesn’t know where to start and feel quite lost.


#2

Hi Francesco,

First of all, thanks for open this thread up, I think the idea to have a more active community is getting more and more important. We’re thinking of new ideas (periodical Licode meetings, Licode days with some hackathon, etc.), but it would be good if we just start with a meeting talking about all these points and know how many people would attend.

We can use this thread to listen from others interested in attending a meeting to talk about the community. We can also define an agenda, schedule the date and decide which communication tool to use.

Regarding VP9 SVC, it’s just a “useless” PoC to know the current state of the feature in Chrome. It would only be usable if you launch Chrome with some special flags, but even in that case we still need to implement the logic in Erizo. Given the need of having to use special flags we wanted to first move the PoC to Simulcast and check whether we can start doing things there before.

Cheers,
Javier.


#3

You guys at Ging are doing a great job with this project.
I also know that some well-know webrtc companies are using Licode, and I have the believe that many companies that are using this project are improving it but didn’t share back sometimes, could be great for the people that really want to help, discuss, document, share and code to have a green path and place to do it so.

Best.


#4

+1, amazing job! :slight_smile:

I also support the idea of having the community more involved, count me in for a meeting talk about the community!

Cheers,


#5

so, are we going in the simulcast vp8 direction?
are all these “add” branches like SenderBitrateEstimation for the simulcast or are there any issue you are trying to address first?
Do you have a roadmap or something in mind?
I don’t want to know the etas, I really don’t care about them.
I’d only love to know what features you have in mind (such as simulcast which would enormously boost the many2many on-demand resolution conference)

btw, I really like the idea of a meeting to discuss about something to keep in touch more easily


#6

yep, Chrome’s Simulcast and SVC are work in progress right now, there are a bunch of Pull Requests related to them, but there are others that fix bugs or that could also be applied to other existing features. For instance, we’re also considering Single Peer Connection for the future. We can’t know the ETA yet because it’s still in a really early development stage, but it has a high priority for us.

Anyway, we’d like to talk about these and other interesting topics in a future meeting with all the people interested in the community. Let me figure out how to organize the meeting and I’ll open a new topic with the info here.


#7

yeah, i saw! I was able to switch from one temporal layer to another with a bit of code and worked like a charm! can’t wait!!


#8

Hi guys,
I bump this thread cause I’ve seen a lot of work done in the past weeks. I’d like to know at where stage is the simulcast feature, it’s a “beta” or there are some fundamental parts still missing?
Love your work,
thanks. Francesco.


#9

I’d say it’s aaaaalmost beta. We’re currently busy implementing a reliable fallback when the available bandwidth for the subscriber is lower than the lowest available layer (https://github.com/lynckia/licode/pull/828) and fixing bugs on the way.
Our idea is to start testing it heavily soon and, after that, make the API public and publish a release in Github. It’s hard to give you a date because we may find more issues to solve but it’s close :slight_smile:


#10

Yeah, I tried the demo and worked flawlessly! Can’t wait to help you test
it in the next weeks. Cheers