We are on our way to make multiplayer for our cyberspace. Since the software isn’t distributed, we plan on having people connect to it using a browser and then use pixel-streaming to send data.
Our machine isn’t that powerful and we do not have the funds to afford services. I have heard that pixel-streaming creates an instance for every user? So we could support like 2 users with our current performance, not good.
The question is, what are the alternatives?
Would it be possible to somehow stream a level, like sending vertex data, collision data etc once to the browser, and then stream only dynamic stuff?
Say we use the BabylonJS framework to display a game made with unreal engine, is that sound? (or sane rather).
Have you found a solution to the problem raised? I am currently struggling with the same problem and am trying to solve it by creating an AWS EC2 for each user, but I don’t think it will be a very pleasant solution.
Look forward to your positive reply.
Thanks, David
Someone needs to do the rendering.
Your options are:
You do the rendering. You pay for the render servers. You have to make sure your business model is profitable enough to sustain this.
The client does the rendering. The player pays for their local machine or console. You hace to make sure that your business model targets players with sufficiently capable machines.
Those are the options. The are no other options. However, you can adjust things like “level of quality needed” to make the client work on as many machines/devices as possible, if your worry is that “Intel Integrated Graphics” won’t be able to render your world.
However, if you don’t want to require installing software on the machine, the only option is “you do the rendering.” For the client machines to do the rendering, they need the software.
Maybe the right choice for you is to target consoles, where the control over software distribution is better. (But then, you instead have to deal with the platform owners and their gatekeeping.)
Hi, Robert. Have you tried using GPU pooling by an ISV - Lark XR, it could empower your content streaming by cluster (s) of GPU on cloud to support multiple cocurrent connected using, plus compatibily with most of the cloud vendors’ auto-scaling automation. that has been widely adopted in Asia and Europe.
Serdar here from Vagon Streams team. If you are still looking for a solution, I would like to share how you can utilize ‘interactive streaming services’ like Vagon Streams, while staying within budget.
As @jwatte mentioned, there are two options for how you can share your application with your users. My post will be mainly about the first option, which you do the rendering process for your users.
Of course, the interactive streaming on cloud resources must need a solid business plan because of the nature of this business. However, you can also keep in budget by using additional limitation features. Here are my personal suggestions for your needs.
Operating System Selection
One of the main drivers of cloud costs is the license costs, so, if your build has Linux support, the costs will be lower than the Windows only builds. At Vagon Streams, we provide affordable prices for both OS.
Choosing the Right Capacity Method
If you will decide to maintain your own cloud resources (ie. EC2) you will have to setup a scalable and also affordable machine orchestration process to always run & stop your streaming machines at the right time. At Vagon Streams, we solved this problem by providing three different capacity methods which can be applicable for all type of use cases by both considering user experience and budget concerns.
Stream Limitations
Unfortunately, one another problem with interactive streaming will be the need to manage your resources according to your user action behaviour. Sometimes, users can leave your streams idle, and cause you some unexpected overcharges on your end too. At Vagon Streams, we provide multiple limitation types which let you to set session limitations according to user behaviours which will help you to keep yourself in budget.
In case you would like to test it, you can check it from here or contact us anytime you want. We will be happy to help.
I didn’t expect to get some replies here after months.
We have not found a solution within UE allthough the offer by Serdar is nice.
For what we need, its not really possible for our target audience. We expect laboratories and schools not owning very powerful machines nor having a great bandwidth.
We decided to reconstruct our project in javascript, then deploy an app on a AWS EC2 instance. This way it’s more lenient for our customers to just open a website and already be inside the game.
