#virtualworlds — Public Fediverse posts

@Metaverse Standards Forum

A few comments by someone with six years of experience in decentralised virtual worlds:

Interesting to see that Sneeze and the Open Metaverse are planned to support glTF PBR, just like Second Life and OpenSim do. Only that PBR has a bad rep on Second Life and OpenSim for bogging viewers down except on what's commonly referred to as "high-end gaming computers". Then again, that's at least partly because PBR-enabled viewers have an advanced lighting model hard-coded that's optional on older, non-PBR viewers.

If Sneeze really wants to be able to render any kind of scene with hundreds of avatars, all decked out in PBR, at a guaranteed constant 60fps on mobile hardware with passive cooling, it'll really need some very advanced cross-platform graphics stack underneath. Although I think that Second Life and OpenSim viewers would be much faster if they managed to switch from early-2000s OpenGL 2.x to modern-day Vulcan.

Granted, one Second Life or OpenSim avatar was more complex with more vertices than an entire World of Warcraft scene some five years ago already, and it has only gotten worse. It'll be interesting how the Open Metaverse is going to keep its avatars lightweight in spite of PBR, and how it's going to keep users from tricking their avatars out to the max. Also, it'll be interesting how the Open Metaverse is going to directly compete with Second Life and its photo-realistic avatars. (Just go search Flickr for "Second Life". All those pics are actual in-world renderings.)

As for Unity and Unreal3D, they're unfit for 3-D virtual worlds because they rely on ready-made scenes. If you want to build in-world rather than pre-build entire scenes in external editors, they're less than optimal to say the least. But virtual worlds live on being built by their users, and they'd better be able to build in-world.

Basic example: Let's say you host a private party, and you need one or two more chairs for your guests. If you have to send them all home, fire up the Unity Editor, open your home scene, edit another two chairs in, edit the corresponding sit scripts in (if needed), save it, export it and import it to the world just to have those extra chairs, that's bad. In Second Life and any OpenSim-based world, you can easily rez another one or two chairs.

But in general, the harder it is to build the world, the less attractive it will be for creative users, and the less the existing users will build it. But the less the world is being built and changed, the more quickly it becomes stale because people have seen everything. One reason why Second Life has survived for more than two decades is because it's constantly changing.

There are some other interesting things I've picked up. For example, content ratings are being considered, including warnings. Not only does that mean that the Open Metaverse is not being designed with the notion in mind that it'll be 100% professional and business-grade, but it means that it'll hopefully have a better content rating system than Second Life and OpenSim, and it'll have a unified definition for the ratings across the whole network, much unlike OpenSim where every grid or even sim owner can redefine "General", "Moderate" and "Adult" as they want to.

As for identities, it's a pity that the Open Metaverse decided to adopt Meta Horizon's concept of digital twins and each human having one and exactly one virtual world identity. Even though this is necessary for security and law enforcement purposes, this is bound to backfire; see this article of mine: Avatars are always digital twins? How about no?

I hope that it'll at least be possible to bring multiple personas online at the same time. This, by the way, would increase persona anonymity: If certain personas are seen in-world frequently, but never in the same place at the same time, this may raise suspicion that there's one and the same person behind them.

(For the record, I've got two fully independent avatars online right now, both attending the same event at the same location.)

This sounds interesting:

Identity does not go offline when a device sleeps, a browser closes, or a network connection drops.

So basically, even if there is no device accessing an identity, the corresponding avatar remains visible in-world?

Of course, this means that it'll have to become part of the culture to send your avatar home or, for those who aren't that much into immersion, at least stash your avatar someplace where it isn't in the way.

What's interesting in the avatar section are the device-specific levels of detail. Essentially, they mean that photo-realistic avatars at least close to the level of Second Life (just look at some pics on Flickr) could be possible, but they most likely won't bog any devices down.

A few things seem not to have been taken into consideration yet: Can avatars change parts of their look (e.g. adding or removing certain accessories or pieces of clothing) in-world in real time? How are high heels for female humanoid avatars handled, especially changing between e.g. high-heeled sandals and barefoot which has to involve adjusting the overall avatar height? How is long hair handled? How are skirts handled?

And yes, the women amongst the users (and probably not only them) will ask for skirts in various lengths and shapes, non-flat shoes and longer-than-above-shoulder, non-ponytail, non-pigtails hair sooner or later. And the skirts had better neither look like concrete nor constantly clip into the legs. (Yes, this is tricky. I've seen about four different takes on skirts in OpenSim. None of them is entirely satisfactory. This really requires inclusion into the very engine as done by Linden Lab's Sansar, High Fidelity in which Philip Rosedale was involved and the two decentralised High Fidelity descendants Vircadia and Overte.)

Lastly, the Voice section in the wiki reminds me of why it's important to solve everything internally rather than licensing external solutions. Otherwise the hassle that Second Life and OpenSim have with finding a working replacement for Vivox is bound to repeat itself.

I've noticed a lack of high-quality media streaming support, e.g. for DJ events or concerts. But seriously, this should really be handled by external media players capable of playing Web streams. Second Life and OpenSim viewers have such stuff built in, but the same stream sounds a lot better through dedicated media players. Just pass the link to the stream on to the users so they can run it through an external media player, and everyone wins.

It really goes to show that creating a satisfactory Metaverse standard is difficult. On the one hand, it really is important to start from scratch and leave behind outdated technology. On the other hand, experiences gained by and in existing worlds must not be disregarded.

#Long #LongPost #CWLong #CWLongPost #SecondLife #OpenSim #OpenSimulator #Metaverse #VirtualWorlds