Recently the CEO of Star Atlas revealed their version of the hangar module, some people were not impressed.
Lumberyard has an advantage in being owned by the people who own Amazon Web Services. Amazon. When you have such an industrial force and presence on the web at your disposal, and one not afraid to attempt new systems, services and methods working for the benefit of what essentially is their own product portfolio integration, it might be harder than one thinks considering the amount of time it has already taken and how long it may still take. This is creating new ways of doing things, it's not picking the best picks from the options on the shelf...yes and no. While unreal 5 does not currently have the asset streaming service as facilitated by CIG (they do allow for streaming assets as does Unity, games like Minecraft stream in the game world as you move around), and hopefully soon a realized server mesh both of these are not new technologies, both Everquest and World of Warcraft have done something similar for years with zones and instances.
SQ42 will have to be that game that makes the waves.I had hoped to have seen Amazon with Lumberyard making waves in the game dev community but so far its not happening
Dont even say it!Unless CR wants to rebuild the game again in UE5 lol
Hopefuly for good reasons and not the Duke Nukem Forever kind of waves. But still the waves I am talking about are what the engine itself can do and not what was created in the engine.SQ42 will have to be that game that makes the waves.
It's true, another engine that is relatively easy to create things in is Unity. Lumberyard is not there yet but I have hope Amazon is going in that direction with it. While I do hope the tools and technologies CIG is crafting on top of Lumberyard will be added to the engine I have a feeling CIG is not going to add all of the modifications and so their codebase is really no longer Lumberyard but a derivative.UE5 is so easy and intuitive to use on the basic surface level, making a demo like this is truly no big deal. You don't need to program anything, you don't need to really know anything special, a very basic surface level understanding of how a map is mad eis enough. If you payed attention to the Cig technical videos of how they build locations, you already have enough knowledge.
You load up a template (UE5 has quiet a few out of the box and you can import community made ones as well) , let's say a first person shooter template for this demo, load in your models which if exported from your 3d modeling software with the correct setting (using the UE exporter plug in) should work fine. But even easier if you just use models from the Megascans library like they did. It's really just some time and no effort.
If you dear leader @Montoya could dedicate a week or two to mess around with UE5, even you could do better then these guys did. It is really that simple, just like these guys proven it is.
I would be curious as to how much time and effort it would take to port SC to UE5. All of the objects, models, and animations should port over quickly. They would be able to throw away all of the LOD's as they are no longer required. As for the planets, it might require some modifications to the engine to be able to read in the planet maps and craft the planets but the porting should be relatively straightforward. Client world streaming is already handled by the open-world partition (https://docs.unrealengine.com/5.0/en-US/world-partition-in-unreal-engine/)On the question of how much better UE5 is to what Cig is having trouble with, no it isn't. They'd run into the same problems with scale, physics, and all the graphical issues they had to solve and so on. They'd have to do it differently, and we might have better performance at the same graphical quality level thanks to systems like nanite and the very well faked GI lighting, but by how much we will never know.
Unless CR wants to rebuild the game again in UE5 lol
(@CR, please don't!)
CIG is still going to have dedicated servers per 100 players or thereabout. It's just the net code instead of being a fixed area will attempt to scale as needed. And honestly, the data is not stored in a database in a typical way as a lookup row/column as that is not efficient. Instead, they are using a tree structure, and that way you just need to grab the node as your parent traverse the tree and get all of the nodes (objects) that exist within. It also gives you the added benefit of being able to quickly add and remove objects from the tree and benefit from a NoSQL DB. I can go into far more detail on networking and data storage but idea is that they are not doing anything untested in regards to server meshing and object container streaming.The net code issues would still be there with UE5, it doesn't have any more or better for what CIG is building.
The current issues with lumberyard stem from not having much to start with and trying to push things like physics grids down its throat. It's vastly more complicated then WoW with its entirely static world and few lines in a database that define everything about the player. Wow had single dedicated servers per x hundred players for a very long time until they switched on the cross server stuff. It clearly takes effort to develop something like this, even when it's rather simple and a huge company like Blizz does it. Now imagine the two net code guys at CIG trying to crack this with no load screens to hide transitions and moving planets with their own gravity and full persistence of every single pice of crap you throw on the ground and such nonsense, all wrapped in a spaghetti code. I imagine 2/3 of the time is just spent migrating everything in the PU into a readable database so they can grab the data from somewhere for all this to happen...
Tldr, you start making a mountain of crap, you gonna have a hard time digging through it later. Thus no true persistence or server meshing for us yet.
Each engine has its strengths and weaknesses, for the longest time Unity was unique in that I could start developing without paying hundreds of thousands for an engine license. I have to say Unreal has been really good for Unity development as well as unlocking a lot of the features that required purchasing to utilize. I really do like that both have opted to go with a percent cut of sales instead of requiring any upfront costs now.I am former Unity developer, the only reason I don't use it now is I don't have the time to work on that kind of project. I also have experience with Unreal and CryEngine. I have no experience with whatever Amazon is doing, I haven't had any time to keep up. The only other engine I saw was a seriously overhyped engine turned out to be a potato. Direct comparisons of those products is like comparing apples, oranges and peaches (and a potato). As a developer, your favorite is going to have very very little to do with anything else other than familiarity. Anyone who thinks there's a use case that isn't covered by one or the other has no idea about how things like inheritance works. Since that is basic to how OOP is written, you just failed the exam.
What I do not understand is some of CIG's internal terminology surrounding this nebulous concept of "server meshing" that has never once been described properly.
The unfortunate thing about that is that since it is obviously undeveloped, they should not be leaking it as if it's equivalent to shard and zone, if it's not shard and zone. Just way too much doublespeak for me to feel comfortable about it.
The servers (as in what is in most user's minds) is a single computer. That is wrong. It is software. In Shard and Zone, the Shard is a service network that allows a number of zones services to share data. If this is what CIG is referring to as "server meshing", then the number of users it is possible to connect to the game at any one time is directly related to the availability of hardware to run the services - and can't ever be infinite, but could be hundreds of thousands of users.
If this is not what "server meshing" is referring to, then CIG is re-inventing the wheel, but so be it.