Yet another Blue Mars thread

I was going to post this as a reply to an earlier thread, but that one seems to have devolved into individual arguments already.

Anyway, I did a little research. And here are some thoughts. . .

You know. . .as it stands now, the only quantifiable differences between BM and SL is the physics engine and I guess the scripting.

From what I understand, now that mono is the engine behind lsl, it might be possible for or other languages to work in SL through mono. Not that I see LL letting that happen--just something that I heard. If someone wants to enlighten me, or point out a flaw in that basic assumption, feel free.

So, next we turn to the physics engine. Havok 4 vs, Cryengine. Does anyone know what version? Last year, Linden Lab updated H1 to H4. I guess updating the grid to H5 is a doable thing. I would assume it would easier than going from H1 to H4.

And then there's prim based building over mesh building. Prims are approachable. But yeah, the flaws and trade offs in this style of building have become more and more apparent over time. But if they wanted to, wouldn't it be possible to let mesh import in? I mean, the grid, and all the prims in it are mesh.

We don't know anything about BM yet other than some technical specifications and hazy semi defined policy from Avatar Reality. Meanwhile, SL is the devil we know. And don't kid yourself--they know who the competition is, and how much competition they actually are.

BTW, doesn't Entropia use CryEngine? they've been around for a few years. There's been no mass exodus from SL.

So, if Blue Mars is that much better, (and we DON'T know that.) couldn't LL just make a switch over from one physics engine to another? Change the avatar mesh? Allow mesh uploads? A better particle engine?

CryEngine isn't a physics engine. It's what is referred to as a "game engine", encompassing all the graphics, the way the world is rendered and presented on your screen as well as movement and collisions and such. Havok is something completely different and comprises only a very small subset of all this. Making SL look nicer and run better isn't a matter of "updating Havok again and adding in some stuff", it's more a matter of "redoing much of SL from scratch to make it work entirely differently".

Mesh uploads would be nice, though it'll give people a lot of bad-fitting skins and clothes all of a sudden, while skin and clothes creators will have to go through great lengths to make their items fit as many meshes as possible, or start stating what meshes their creations are intended for.

Then of course there's the thing about the appearance/shape editor. That thing will be affected by mesh changes as well, custom meshes making it even harder to handle things. I'd love to see new meshes, maybe even import some like the ones used at Renderosity, but it'll be hard to get things like that sorted.

Another point is that prims are not meshes.

A prim is a "geometric primitive" It's defined by a set of mathematical equations, e.g. the surface of a sphere is defined as being at radius r from its center. If it's a cut sphere, additional equations define the parts to be excluded, and so on.

This is complex, but it's a LOT less complex than a mesh, in which a large number of points and their relative positions must be individually dealt with.

The only "meshes" I'm aware of in SL are the avatar mesh and the land mesh, both of which are relatively coarse.

Sculpties are a sort of pseudo-mesh. The sculpt map serves to define the point locations...but the calculations are all handed off to the user's computer, which is why sculpties act like their base sphere in terms of SL physics...the server doesn't "see" their end shape.

Which sort of gives you an idea of what transitioning SL to a mesh-based model would entail. Imagine the render times if EVERYTHING were a sculpty.

(Now somebody, probably Argent, will pop in here and tell me I don't know what I'm talking about, which is probably true. The inner workings of 3D software of any type are largely a mystery to me.)

And the fact that not everyone can create or script, and the fact that not all content will be 100% functional in all 'worlds', and the fact that the System Specs for computers running it will be higher, and well, I could keep going on. There are many quantifiable differences, So many that I can't rightly place the two in the same class. BM is not a world, but better s showcase of realms, or rooms if you will. It's closer to the IMVU model than the SL model.

Not quite correct. Everything comes down to meshes. The difference is that, instead of a static model mesh data block being used to make a primitive, a set of parametric math equations are used to generate it on-the-fly. The upsides: no storage required to hold the mesh (except for the prim parameters) until it is ready to be rendered; this also means, in theory, a lot smaller data transfer when SL streams the scene to the viewer. The downside: generating complex meshes for primitives, lots of them, takes time during the rendering phase, so the viewer will lag sooner as the scene complexity grows. Another downside is that you can't easily represent a complex surface exactly with prims. You can get "good enough", but often that takes a LOT of prims. Sculpties were meant to alleviate this problem, but there are many shapes which are difficult to simulate with sculpties.

In the end, the generated SL scene is just as complex, triangle-wise, as ones generated in many games; often, even more complex. Meshes do allow much better control of a developer's "polygon budget" but, just like particles, can easily be badly abused by the neophyte.

Some may argue that the advantage in data transfer which prims have over meshes is important. As I said above, it is, in theory, an upside. In actual practice, I don't believe it is much of an upside, if one at all. Compared to the number of primitives used to construct something complex, like a Dragon avatar, a mesh would likely take up less bandwidth. The same is true for the scenes themselves.

I will say that, to properly take advantage of meshes does require about an order of magnitude more skill than with prims, so prims definitely win out in the "ease of use" category.



In actuality, a sculpty has no more triangles in it than a torus. Since people often go crazy with toruses in their builds, you get about the same effect if they were all sculpties. However, being all-sculpty means you wouldn't have to use near as many, so there is a significant savings available.

Just like I mentioned, meshes don't necessarily mean more lag. You most definitely can abuse meshes like you can abuse prims. However, if you know what you are doing, you can actually make most scenes simpler and more accurate with meshes than you can with prims, even sculpties.



Well, you're not far off. Just a few intimate details that needed to be cleaned up.

Talarus hit the nail on the head, there. IF used correctly, mesh import would be no more polygon heavy than prim creation. In fact, for more complex builds, meshes are more economical. IF used correctly.

If, if, if.

The thing about prims in SecondLife is that for basic building, they encourage intelligent use of polygons. Want to make a build that is basically so many boxes? You don't need thousands of polygons to do that.

This all goes out the proverbial window if one starts using gobs of toruses and sculpties, however. And even judicious sculpt use can add up the polygons faster than a nice, simple mesh would.

Thanks Anti for that clarification.

But, the fact that SL is composed of bits and pieces--a scripting engine, particle engine, physics engine, etc and so forth. . .Doesn't that make SL a little more adaptable? I mean, if a new CryEngine comes out, there's no law that says it has to be compatible with everything that came before. Likely, there probably WOULD be an inherent level of compatibility, but, well, when LL upgraded Havok, that was a painful process and it was only the physics engine. If a new CryEngine was released, Avatar Reality would have to update the whole thing, (or run the risk of "looking old" and I imagine that would be a considerably more involved task.

As for Linden Lab being reluctant to to make changes to avatar mesh, or the way things are created, scripted, illuminated, moved, paid for or any other aspect of Second life--We ARE all talking about the same company that's sculpties, mono, windlight, havok 4 and eliminated wire transfers? I get the impression that the powers that be in San Fransisco will do what they damn please to the grid.

As for me assuming that "prims are mesh" uh. . .while I don't understand the details as well as Talarus, It's easy enough to drop into wireframe view and see that even prims are composed of triangles. Calculated vs stored is the sort of detail I was hazy on.

Well, it is kind of a different situation. Technically, CryEngine is a "game engine" as has been said. Developers usually take one version of it and make a complete, standalone game with it. They don't have a need to change the "engine" or any of its parts on the fly for that product, unlike SL. They will just make a new game based on a newer engine when it comes out.

SL is a persistent world, which means that the only way it can change is "on the fly". They can't easily just develop SL 2.0, get everyone to migrate over, then shut off SL 1.0. People have a lot of investment in SL 1.0 and, likely, a lot of it wouldn't transfer to SL 2.0.

As for the situation with Avatar Reality and CryEngine, I don't know how much of it they used, and what their upgrade strategy is. I doubt it will be less painful than doing the same thing with SL, however. It is simply the nature of technological advancement, especially in the game / entertainment industry.



Oh, that is a given; that LL will do whatever it thinks is in SL's best interests (read: whatever it damn well wants, to hell with the Residents).

As long as we keep upgrading is good, I really would have liked to see some statements from LLforcasting next years new features like Havok 5 or Avatar mesh, puppeteering etc.
They can't change too much too quickly because people get upset when they suddenly find the product they have been selling since 2004 is outdated.

The Opensim devs said exactly the same thing.

So if a radical SL 2.0 is out of the question, just what could they do with SL 1.0 to improve things and respond, at least partly, to the technical superiority of not just Blue Mars, but every other erstwhile competitor that comes along, using whatever is the very latest in 3D advances?

The life left in SL will be determined not by its existing resident population, but by its ability to attract and retain new residents, and for them, first impressions will be key. Are we really going to see an aging SL, populated by an aging diehard fanbase, both sinking into the sunset together? What can SL do to attract new users, and keep them?

We have an idea of the numbers of people who hear about SL, create an account, visit it once or twice, then never return again. It will be interesting to see if BM have the same sort of low retention figures. But if BM have higher retention figures, then other competitors will also benefit, after analysing the reasons.

Interesting times ahead, for sure.

Rock

With the importance sculpts have gained, SL pretty much *is* mesh-based. I don't think I've bought anything inworld for the last year which didn't include *some* sculpts.

Sculpts are just a horribly convoluted, not terribly effective, and imprecise way of describing meshes.

Honestly, I think sculpts was a mistake. You need an external 3D modeling tool to generate them anyway, and *on top* of that, you need to jump through horribly convoluted hoops to import them with any accuracy. The only gain is the compression when transferring the vertex data to the client, but most people use either uncompressed sculpt maps or are doing hacks and workarounds, using larger-than-necessary sculpts to control the edges.

The same what it did to us?!

I think, the trick of SL was and is, to deliver nothing. It keeps mostly people who are somehow creative (without judging the creations). It starts with landing as ugly nothing. *If* one is logging in again, a motivation for doing some self-design and self-expression appears. At first by buying things, later by do-it-yourself.

BM is pre-formatted, pre-configured. A museum. Or, if there is any action, then only like booked holidays where all is done by the travel-agency and the hotel.

This looks nice at first sight, but very soon it will become boring. The remarkable aspect of a museum is, that it is a container for dead things. To visit a museum gives a hint to the visitors, how it may feel to watch a grave from inside. The best thing of a museum is the exit and the cataloque-stand and to see the active world outside again, including a coffee and a nice snack.

To do and to create something and to kill it at a specific point of boredom and to do and to create something new, maybe then on a better skilled level again, is the trick of SL.

This is not made for too passive and lethargic consumer attitudes. It is made for being - somehow - constantly active.

People with a passive mentality, trained by life-long passive media consumption from youth on, are lost in SL.

It is just not made for everyone.

BM is Disneyland par excellence. All is cloned and from a specific point on, you cannot be sure if a human made some scenes and builds there or a program...

SL lives by human life, creative people and endless changes.

If that stops, forced by some crazy ideas of the actual management..., then is SL as dead as any museum, where visitors can change nothing. It would become BM...

The original framing story was that Blue Mars is, indeed, Mars terraformed as a sort of holiday resort/created paradise, in the year 2177.
Though how much that story actually still holds up, and if it has any practical impact at all, I don't know. Still, I can't help but think of Blue Mars as Phloston Paradise

For SL to stay relevant, I think it needs to focus on being better at what it was originally intended as: A platform for people to make their own things in. It needs to decentralize control; not gather it under one "Linden Rule of Predictability and Disneyfication". It seems to me that LL is trying to turn SL into something which can compete with the spin Blue Mars is taking, rather than believing in what they already have (or had).

I think tho, what I'm actually trying to understand, or decide, is whether or not SL can keep the technology interesting. There was a time when there was no flex, no scuplts, no voice. and the particle illumination/weather system was a lot simpler. They keep tweaking and modifying the infrastructure and coming up with new features. They've changed policy left and right to (grudgingly) comply with laws and even (to a point) resident expectations.

But I'm mostly interested in the technology. Does anyone even know what the inherent limitations in the conglomeration of technologies we call the grid? What can be changed, what can't? Can mesh coexist with prims? Does a new avatar mesh HAVE to destroy all the clothing that came before? Does the current architecture mean I'll never step from one sim to the next without zipping across to a third?

Can mesh coexist with prims? Yes Prims are meshes, just pre-defined.

Does a new avatar mesh HAVE to destroy all the clothing that came before?

Not necessarily. But it would require the original creator to re-factor the old clothing to the new avatar mesh, leaving those who bought out of luck. Or, they could make which mesh is used a user-selectable option, so you could toggle between the two.

Does the current architecture mean I'll never step from one sim to the next without zipping across to a third?

I think it's possible to overcome this, but I suspect the cost of the hardware upgrades to do it are probably prohibitive.

I mentioned WoW the other day, and I will again, only because I think it illustrates a point. WoW hasn't made any substantial upgrades to their graphics engine in over 5 years. It's still extremely primitive and cartoonish, even when compared to SL. Yet, it's still one of the most popular online games. I think this points to something of key importance, The average user isn't that concerned with how wonderful and state of the art the graphics are. What they are concerned with is the functionality, the activities, and the social aspect.

WoW works well, there's a lot to do there, and people make a lot of friends.

SL has a good handle on the social, but the other two are where it needs improvement. Rating a platform on how often they upgrade their graphics is a bit short sighted in my opinion.

I hope not. I'd hate to see them go through what happened to Palm when they decided that instead of making great *organizers* they got sucked into competing with Microsoft and trying to make laptop replacements.

There's still a notable difference between WoW & SL

WoW doesnt change landscapes daily. Everything is already made for you. Nothing is really "new". Its already on your hard drive. All that needs transferred on the network are movements, maybe a few other things.

Technology-wise, it succeeded because it really doesnt have demanding graphic requirements.

Warcraft System Requirements:
Processor:
Minimum: Intel Pentium 4 1.3 GHz or AMD Athlon XP 1500+

Memory:
Minimum: 512 MB RAM (1GB for Vista users)
Recommended: 1 GB RAM (2 GB for Vista users)

Video:
Minimum: 3D graphics processor with Hardware Transform and Lighting with 32 MB VRAM Such as an ATI Radeon 7200 or NVIDIA GeForce 2 class card or better
Recommended: 3D graphics processor with Vertex and Pixel Shader capability with 128 MB VRAM Such as an ATI Radeon X1600 or NVIDIA GeForce 7600 GT class card or better

Those are pretty low standards for "Minimum" which is how it keeps 10 Million+ users.

I'd like to see SL run on a GeForce 2

And anything with the CryEngine? LOL!!!! Forget it!

I can understand the "why" of BM Creator rules. Would anyone care to try streaming a 100,000,000 polygon mesh thats moving in real time?



For the company- yes. For the end user, not really.

Wouldn't it be possible to have the new mesh use a UV map approximating the wrapping on the old mesh when used with old textures/clothes?
-It would probably look a little funky here and there, but likely not all that much more than where the mesh weirds out today anyway.

I don't think that will work anymore. When SL started out it was unique, and had no direct competition, and that has continued to be the case, until BM came along.

If BM does launch, then I think that the hype around BM will ensure that most people who are thinking of looking at this virtual world stuff will look at both SL and BM.

If that turns out to be the case, and bearing in mind that first impressions are so important, why would would a visitor to SL choose it over BM to come back for a second look?

As for in-world creation tools (and there really is only prim creation and basic vegetation in SL) I doubt that a significant percentage of newcomers to SL or BM create anything on their first day. Most (I believe) will explore first.

If I was LL I would make sure that Orientation Island and the Welcome areas were mini-continents, and I would put there the very best that SL had to offer. I would still be nervous how that would stack up with the best that BM had to offer.

Rock

It's possible, but ending up with something funky looking is kind of against the whole idea of a better mesh with better UV mapping. If the end result isn't substantially better looking, may as well not do it at all.

I probably should have said toggling between the old and new UV mapping, that would also be workable.

Freedom.

Freedom Fries!

Can you explain that in more detail. A newcomer spends an hour in SL, then spends an hour in BM, having a look around. What freedom, precisely, would they notice in SL?

Rock

Mesh can coexist with prims. Realxtend have a system already developed for that and it currently working to make it part of Opensim.
A new avatar mesh don't have to, but a more practical way would be to allow both, the legacy avatar and new avatar mesh system.