Issue with pixel streaming server's configuration

[Muzaheed (ahsan Muzaheed)]

@NC3D

Colin,

Thanks so much for the demo! We finally got our site working on Azure after restoring the original config.json (the one that Epic says to edit) and putting our server details in the AWS Powershell script.

We want to have many people able to view a single instance of the game. I had thought that the TURN server is supposed to re-broadcast the stream to handle multiple connections, but we always end up with very blocky images and strange compression glitches when we get more than a few clients on our server – even when we’re only running the included TURN server on Azure.

Has anybody had success with multiple connections to the same game? We’re having a very hard time telling whether the TURN server is doing its job: we’re getting around routers, but we don’t see any other signs that the server is doing anything. I thought that with TURN, the game will send one stream at the rate specified and the TURN server will re-broadcast that same stream to as many clients as it can…?

-Donald

u will need a media server. which will broacdcast same video stream to all connected user.
do googling about webrtc media sever .

@smudgy67

Hey Colin,

Thank you for the reply and for confirming what AWS stands for.
In relation to “I thought the idea of pixel streaming was to use the graphics capabilities of the host machine, which is why I’m trying to use mine.”, is this not correct too?!
Seems AWS has varying levels at what graphical hardware is used to deliver the PS service. Again, this points me back to wanting to use my own machine to deliver the service.

I got to the point where my friend connected to me across the internet via the player.htm page, on hitting the play button, the server would not connect. He could access the menu and managed to kick an instance of the app on my localhost.

I have now got a free developer account with xirsys.com, they provide STUN/TURN server access.

After looking through Epic’s documentation, they gear this round AWS connectivity, however. it would be nice to know how to set this stuff up with alternate suppliers.
I know you think my approach is a headache, however, if my method eventually works, it means other developers out there can use the PS service with the best possible results and manage this with a low budget too. This means it frees them up to develop UE4 applications and not get lost in the mechanics of a service that should work just out of the box (within reason).

I appreciate all the help you have provided on this topic, however, if you could assist further, I’d really appreciate it, thank you.

All the best,

Smudgy (smudgy67@msn.com)

​​​​​​​

i am following that plan. we have 10 computer with powerful gpu in our office behind NAT.
we are using those GPU’s unused power to server to user who want to test out app in pixel straming.
so far it worked like a dream.

and video qulaity is super as each intance is using single powerul gpu.
and the user who using the computer have no idea that his computer also serving as a GPU cloud for pixel straming