That's interesting. It seems to be a system where you create an area in Unity and upload it, after which others can visit and go to it via portals. Not clear how much asset-sharing and moving around is supported.

This seems to be like SineSpace, Sansar, HighFidelity, etc. I call those “game level loaders”. (Is there a more accepted name?) People can make a level and upload it. Most seem to use Unity for creation. The server hosts it, and it works mostly by downloading the entire level to the client. Going through a portal means downloading a new level, which takes a while.

Some have more sharing than others. In some, you can upload an area onto a land parcel you own, and you can see and visit the adjacent parcels of others. Decentraland and Sominium Space seem to work that way.

This is a way to get something going without tackling the big shared world problem head on. However, none of the game level loader systems seem to be that successful.

Well From my understanding…(If I may hypothesize) I've been studying these platforms for about 4 years. Somnium space created a single Unity scene which is a “complete world” it takes about 6 GB to download currently (with mostly empty parcels).

Each parcel is essentially an compressed asset pack, so you can expect that single world to balloon in size if people actually were to populate it. It's interesting for sure, but not well partitioned. It simply won't scale. In addition it's limited to basic graphical models/animations. It can't run

code against those parcels, because your parcel's code could corrupt mine and crash the experience etc. They're trying to overcome said limitation with “sominum worlds” and so on.

I come from a background in enterprise SaaS architecture, where knowing what's going to stop you is something you avoid in design. We deal in realisms. So we did not take that approach. Realistically, partitioning is important if you want a vast metaverse. If we were talking about the web, no one would

expect you to download EVERY website to experience the web. That's ridiculous ?.

NeosVR is another unique animal, allowing spawns from inventory across worlds. I really like it. I would say I'm a fan. I think it has an great future. But ultimately I believe it will suffer when it comes to general game design and their difficulty with physics/ Rigidbody's seems to back that up.

Our goal is to create a metaverse as one expects, as Baszuki expects. One that is not limited to just social uses like chat or concerts, but to build full multiplayer games. We started with visual node scripting like Neos, realized its limitations, our research led us elsewhere.

What we have (while early) is a working, scalable platform that incorporates the strength of modern game engines, provides a secure experience for the users, and provides back end services necessary to provide a full economy, marketplace etc.

I'm excited to see how it evolves. It's a zero hype project. We're not pushing blockchain (although you can link your NFTs if you choose…totally up to the developer).

We're singularly focused on the metaverse itself. It will be a great place for content creators to build what they want and do so without “paying for virtual land” and all that garbage. It's early, but at least the potential is there. Cool stuff.

Regarding “game level loader systems”..it's still pretty early. If we consider the big shared world problem: 1000 users and 1000 objects is 1000x1000 synchronization calls. So, that's not happening anytime soon. Our goal is to start with a metaverse that works and optimize it,

not promise a borderless system that would never work (talking to you E2 ?…j/k they don't hear me cause none of them are devs…lol).

MassiveLoop_Metaverse said:
“not promise a borderless system that would never work (talking to you E2 ?…j/k they don't hear me cause none of them are devs…lol).”

Um, yes. Most of the NFT crowd has either a very basic 3D world, or no 3D world at all.

I like big seamless worlds, and, as I've mentioned, am writing a Second Life viewer in Rust. Turns out you can get 60FPS display on a gamer machine by using Rust, WGPU, Vulkan, lots of threads, and effective asset loading which changes with the viewpoint to keep the right stuff in VRAM. The architecture isn't that bad. It's the 20 year old single thread OpenGL C++ implementation that's the problem.

It turns out the real problem with big, dynamic, user-created worlds, is that once they really get going, the content statistics look nothing like typical game content. Since you have a competitive market in content, people compete by making high-detail, nice looking content. If you have complexity limits, people will game them. There's very little instancing, since each creator creates their own content. So you're sucking huge amounts of bandwidth from the asset servers. (If the NFT people ever get serious user counts in a 3D world, this will hit them even harder, since the whole point of NFTs is to have only one instance of the NFT.)

The other content problem is when things get optimized. In most games, groups of objects and avatar layers of clothing are all ground down to simpler models with precomputed LODs during level development. In a more dynamic environment, you have to do that at “run time”. Second Life does not do that very efficiently, and does too much unnecessary drawing. This is the source of the claim that worlds with user-created content have to be either slow or low-detail. That's not true, but it does require doing things somewhat differently. Roblox is working on a new, impressive clothing system which does “baking” when you change clothes. “Run time” can also be “clothing change time”. See their dev blog.

These are some of the problems you hit as these systems scale up. So far, Roblox and Second Life are the main systems that got far enough to hit them. You have all the problems of an MMO, plus these. They're quite solvable; they just haven't had enough attention yet.

That sounds like a very complex project (SL viewer). SL is pretty amazing, the real pioneers. It is surprising to me that they turned over Sansar development.

I think partitioning is important. The level of partitioning will decrease over time with Nielsen's law behind you, but it's always going to be important. A metaverse is a like a universe. You can have a big open world, but it would

still only be one world among many. There would never be “enough” speed or space anymore than there is enough to deliver HTTP content today. We just keep finding new ways to make the client experience richer and that costs more bandwidth.

NFTs…I do agree. It's hard to imagine it being practical to validate every unique sword or shield against the blockchain directly without it being the slowest game ever played. Still young tech, so I'm sure it will improve, but

there's really no reason to run back to the blockchain unless those items are “cross game” or “cross platform”. I mean if they're just in a single game, our game logic must choose to respect the role of those NFTs or not.

So the blockchain is subject to the game logic anyway!?!. So, I don't know. Maybe if they're used across games and there is an abstracted 3rd party system that validates the items. But if the NFTs functions are all completely encapsulated

into my own single game, (call me crazy), but I think a database would have worked just as well (with way less network latency) since the blockchain is not the real authority, the game is.

MassiveLoop_Metaverse said:
That sounds like a very complex project (SL viewer). SL is pretty amazing, the real pioneers.

Yes. Second Life already has AAA-title visual quality in the content, but you can't see it clearly because the client, which is single-thread OpenGL, is so sluggish. So I've been working on that. Here's some recent video:

This is a demo of playing existing Second Life content with a more modern and much faster viewer. Rust, Vulkan, WGPU, PBR, multiple threads, better prioritization of what to load, etc. The viewer is frantically moving textures from the servers to the disk cache to the GPU, switching LODs to prevent over-filling VRAM while trying to stay ahead of the camera movement. Refresh rate was about 54 FPS.

(This is all very experimental. This connects to Second Life servers and moves the camera, but you can't interact with the world yet. Focusing on the rendering end for now. No avatars yet, because WGPU doesn't have rigged mesh yet.)

So, metaverses do not have to be low-rez. They can come up to the GTA-V level, at least.

Roblox gets this; see their development roadmap.

You need a gamer-level client machine, at least what the average Steam user has, which is an NVidia 1080.

Well, Zuckerberg announced the name change from “Facebook” to “Meta”.

This was not accompanied by any product announcement. Facebook's three tries at virtual worlds have been flops. But Zuckerberg is now being called the “mayor of the metaverse”.

The ratio of hype to working systems in this area is excessive.

Nagle said:
They can come up to the GTA-V level, at least.

I don't think they come, for the reasons that you suggested in a previous thread: Everyone has a custom thing.

In GTA, you will note that, as you drive around, there are “themes” to the vehicles that show up. For a while, there are lots of hatchbacks, then there's lots of sports cars, then there's lots of minivans ….

I believe this is because they re-use the meshes and just re-skin them, and only have a fixed number of meshes loaded. Sometimes, they unload one of the N meshes, and load another one in its place, and the character of cars changes. This lets them have more vehicles for less loaded RAM. Similarly, there's a fair number of “color re-skins” where normal maps and such are re-used, but the color is tinted/changed for the vehicle.

In the metaverse, though, there aren't just 30 different vehicle meshes to choose from; everyone wants to make and sell their own mesh, and on a typical road, almost every single mesh will be custom.

@Tom Sloper @nagle

I agree, ratio of hype to working systems is crazy. I have seen projects that literally just post advertisements of Unreal engine on their site and people automatically believe it is a metaverse. Even though they have years of bugs and networking issues to work out.

We do at least have an honest working product . Far more customizable than Horizon, and better looking as well (sample halloween post below). But it seems to me he's lashing out somewhat trying to deter other entrants…Not deterred, not impressed…a bit entertained.

@Nagle Yes that is very cool. The textures look quite good. It would be so cool to really walk around it. Can you build VR into your viewer even though they are not?