Star Citizen Dev Progress Watch

[Judge_dolly_OG]

Definitely not but I trust boredgamer and his unbiased SC intuition for December release.

"Welcome to some more Star Citizen looking at the new Roadmap and Focusing on Star Citizen Alpha 3.4 which will be release around the end of December, while it’s testing phase is expected to start by the end of November."

It's quite clear now that 3.5 is going to be THE patch

[jwh1701]

It's quite clear now that 3.5 is going to be THE patch

For fun I have been telling anyone that 3.6 is the patch to wait for on reddit.

[dsmart]

CIG: So incompetent that even when they are putting the player on rails in an animation that is literally rails they can't even stay on the rails.

They can't even get vertical elevators to work, and you think they can get a lateral one to work?

[dsmart]

I do not understand how bored gets up everyday and keeps making SC videos with so much belief. If he's trying to make money on yt SC is not doing it for him.

Never stop shilling is the motto.

[Motto]

Never stop shilling is the motto.

I have nothing to do with shilling

[dsmart]

In the latest FTR video, I made this post about Star Citizen instancing. I am posting it here for bookmarking.

Chris Roberts verbal diarrhea, circa Nov 2012.

It's already instanced. Each 50 player server is an instance of the game world that runs a copy of it spooled up via AWS. The "server meshing" which they keep talking about, is the connection of several instances to give the illusion of a large cohesive universe. In order to do that, they have two options:

1) Connect instances together so that players in Instance1, can move to Instance2 and vice-versa. This is trivial but not feasible because there is no point in having intra-system travel if it's the same world. How much sense would it make for a player to move from STANTON in Instance1 to STANTON in Instance2?

2) Break up the entire game world so that each star system is an instance. In this manner, STANTON and KAYFA would be their own instances. Doing it this way, players in STANTON instance would move to KAYFA instance and vice-versa. This would be their inter-system travel. That would work - except for the part where they are still stuck at 50 players per server instance where the server literally DIES during periods where more than 12 clients are doing "gameplay" things (e.g. space combat, fps combat, fast travel, mining, EVA etc). So if they do this - which is precisely what I believe they would do IF the project gets that far - they would still be stuck with 50 client instances because their networking layer, engine, and world structure, aren't conducive to an MMO. If you build 20 star systems with 50 players each, and they are all logged in at the same time, well that's your 1000 player world right there. The hilarious part is that they haven't even finished building a SINGLE star system (Stanton) yet. They promised 106 of them back when they raised $65M in Nov 2014.

In the CitizenCon clip which you played, I believe #2 is precisely what croberts is describing. Note how he carefully worded the "1000 players" statement, which is a complete walkback of previous statements by him, Erin, and others in the dev team about server meshing supporting thousands of players in the same area. That's just never - ever going to happen. Not even the venerable Planetside2 could do it. Even if they lose their minds and implemented something like SpatialOS (which would be a major undertaking given that it doesn't have native support for CryEngine or Lumberyard, and requires MASSIVE changes to the pre-existing custom game engine and world space), they would barely be able to retain 100 clients per instance at the current fidelity, networking kernel etc.

For now until eternity, unless they do a major overhaul (note that the OSC + NBC overhaul is now 18+ months and still not finished, nor yielded the expected results) of the game's underlying architecture - a massive undertaking seeing that it's already 6 yrs in dev - it's going to remain a 50 client instanced game. Even if they pulled off server meshing as I described above.

FYI Elite Dangerous does this with dynamically allocated instances (they call them islands) in a 32 client P2P environment; and it can grow to about 100 - with disastrous results (read up on what happened in Summer 2017 when about 3000 players engaged in a massive battle). But they never claimed to be making an MMO. So.

croberts talking about server meshing, CitizenCon 2018 @ 14:44

[N0mad]

I bet Chris is kicking himself for not using Unreal. If he had then they would now have access to Spatial OS. But they can't use it with Lumberyard because they're contractually locked in to Amazon's server systems, which don't currently support it - LOL.

[jwh1701]

I bet Chris is kicking himself for not using Unreal. If he had then they would now have access to Spatial OS. But they can't use it with Lumberyard because they're contractually locked in to Amazon's server systems, which don't currently support it - LOL.

The spatial videos from unity and unreal are really interesting, but I wonder was this very difficult to develop SP OS? I wonder why unity or unreal would not have done something similar but directly encoded into the engine for easier development?

[N0mad]

The spatial videos from unity and unreal are really interesting, but I wonder was this very difficult to develop SP OS? I wonder why unity or unreal would not have done something similar but directly encoded into the engine for easier development?

It is hard. That's not to say it hasn't been done before with varying success: EVE online, Planetside 2, and more recently Dual Universe. The difference with Spatial OS is that it's a service for any game - and (although I haven't seen the details) they effectively have integrated with Unity and Unreal. All the complexities of netcode and server meshing are hidden away behind the API.

Basically, it's exactly what Chris wants, but CIG are stuck trying develop their own custom version by themselves. They just can't. And since the deal with Amazon Lumberyard locks them into using Amazon's server infrastructure there's no way to use Spatial OS (because you have to use the Spatial OS servers - which are likely part of Google's infrastructure). Amazon could develop a server meshing system of their own, but that'll be years away, if ever.

[jwh1701]

I have nothing to do with shilling

LOL I have always suspected you were shilling and subtle hint was I all I needed.

[Motto]

Even if I was, I absolve myself of that 

[Judge_dolly_OG]

The spatial videos from unity and unreal are really interesting, but I wonder was this very difficult to develop SP OS? I wonder why unity or unreal would not have done something similar but directly encoded into the engine for easier development?

I wonder how long it took to develop this tech, how many years away would CIG be from being able to implement something this, if they managed to get the necessary expertise and funds to do it?

Given that they are starting with a game engine that is unsuitable and already creaking under the strain, and need to keep current backers funneling money in so have to show signs of progress in the general game, then they must be looking at about five years at absolute best case scenario? If indeed it is possible at all on AWS.

I'm guessing that what they are actually going to try and implement won't be anything like SpatialOS level stuff.

[dsmart]

I wonder how long it took to develop this tech, how many years away would CIG be from being able to implement something this, if they managed to get the necessary expertise and funds to do it?

Given that they are starting with a game engine that is unsuitable and already creaking under the strain, and need to keep current backers funneling money in so have to show signs of progress in the general game, then they must be looking at about five years at absolute best case scenario? If indeed it is possible at all on AWS.

I'm guessing that what they are actually going to try and implement won't be anything like SpatialOS level stuff.

Aside from the fact that they have their own underlying world builder and multiplayer backend built with CryEngine/Lumberyard, moving to SpatialOS won't yield any benefits for Star Citizen.

Also, it would be a LOT more difficult for them to migrate to SpatialOS because it only has integrated support (via plugins) for UE4 and Unity. Which means that they would have to do a lot more work to port it to their CryEngine/Lumberyard/StarEngine soup.

And it doesn't solve their networking problems because that's not what SpatialOS does. It's just a server backend. You still have to bring your engine, networking layer etc.

As I wrote in another post, their only option to increase the size of their world and implement inter-system travel, is by making each star system it's own instance. What's going to be funny is with their shitty networking, there's a good chance that jumping from one system to another is probably going to break, disconnect etc. God it's going to be so hilarious. Assuming they ever get that far. Seeing as the current Q2/19 dev schedule doesn't even have anything outside of Stanton, it's highly unlikely that they're going to do any of that in 2019.

[Judge_dolly_OG]

Aside from the fact that they have their own underlying world builder and multiplayer backend built with CryEngine/Lumberyard, moving to SpatialOS won't yield any benefits for Star Citizen.

Also, it would be a LOT more difficult for them to migrate to SpatialOS because it only has integrated support (via plugins) for UE4 and Unity. Which means that they would have to do a lot more work to port it to their CryEngine/Lumberyard/StarEngine soup.

And it doesn't solve their networking problems because that's not what SpatialOS does. It's just a server backend. You still have to bring your engine, networking layer etc.

As I wrote in another post, their only option to increase the size of their world and implement inter-system travel, is by making each star system it's own instance. What's going to be funny is with their shitty networking, there's a good chance that jumping from one system to another is probably going to break, disconnect etc. God it's going to be so hilarious. Assuming they ever get that far. Seeing as the current Q2/19 dev schedule doesn't even have anything outside of Stanton, it's highly unlikely that they're going to do any of that in 2019.

I was wondering about them doing their own implementation of something like this, to enable those 1000 player space battles they were talking about. Though I think I saw something about them walking back from that statement recently to be honest.

Would having one server per system actually improve anything anyway? They are basically doing that now (without even a full system).

[jwh1701]

engine, networking layer etc

Interesting they way the showed off the tech and talked about I though it was an all in one solution to ease the burden
on anyone wanting to make a massive MMO. Since they cannot even get the basics down I do not see them getting the engine
or networking to a respectable level.