Yet another Blue Mars thread

So, most of you seem to think that SL is never going to be able to overcome certain things like the bad mesh, or grid crossings without a large cost to Linden Lab, or, destroying most existing content, or both.

Is that written in stone? I mean, let's take grid crossings, for example. As I understand it, the grid is actually a bunch of interconnected databases that define what things look like and where they are, and what they're doing at any given time.

Is there no way to, for example find a way to weight the data pertaining to things that are likely to cross borders, vs things that never seem to move, or, simplify/compress/expedite (or whatever) the data that moves between sims? What exactly IS the bottleneck?

Actually, it is normal in many games to break up large surfaces into significantly more triangles than needed, especially large, flat surfaces. Why? Local lighting. Local lighting works at the triangle level. The less triangles you have, the more unnatural the lighting of them looks.

As for hidden polygons, you're right; they are a waste.



I dunno. While it may be wasteful, some of the builds with prims are simply amazing. Svarga, for example. They most certainly don't look like they are made out of "Legos".

Anything is possible; it just comes down to a matter of time. In this case, more time than LL can afford to spend, or even has in terms of its present amount of manpower.



Almost. More accurately, the grid is actually a bunch of interconnected, isolated simulations. When you move from one simulation to another, your entire state has to be transferred and synchronized between the two. While that is happening, the simulation you are leaving has to stop simulating you, and the one you are entering has to wait until it has your full state to start simulating you. If the simulations can do this at their leisure, synchronously, and only when absolutely necessary, then it will appear seamless. If they HAVE to do it RIGHT DAMN NOW, which is what hard boundaries force them to do, then it will appear laggy with rubberbanding artifacts.



The bottleneck is simply how they go about it, and it is based on some design decisions which are at the core of the SL grid "engine". Thus, changing them would most likely have a nasty ripple effect across many parts of the SL "world", not the least of which is the content. For example, many people have designed their products to depend on the hard-boundary nature of the grid simulations. Making those boundaries flexible, or removing them altogether, would mean those dependencies would become invalid, and their products would break.

Personally, I take breakage in the name of progress as a fact of life. Good content developers will take them in stride, both by designing their products to be open to such transitions, and updating them regularly, quickly, and easily when they happen. However, many folks are simply resistant to change, digging their heels and saying "no way!".

*sighs* Rock, with some things, yes, they may develop over time.

Fact is though that is has been stated in the interview linked earlier that you cannot switch your gender once you defined it. Fact is, so far I can't wear hair or clothing that is marked for a character of the opposite gender (and I have no reason to believe this will be different in other cities since this seems rather like a technical decision). And fact is, they haven't decided on the sex yet (see also same linked interview)... and given their big name investors, I highly doubt they'll go for it and sully their brilliant reputation *coughs* by the nasty cyber sex association.

That said, you are a pompous dickhead that strangely only knows all in these forums, and I am too tired for this crap. Can I just have a snapshot when they tell you no sexes and you have to kiss your next place of stylish submission goodbye? Thnx.

At any one given time, the sim you are standing in is loaded, as well as the 4 sims surrounding you. In SL, these crossings are only partially threaded according to draw distance. Crossing a region is not just a matter of a simple handoff to the next CPU. It's is a matter of (a) dumping the old data, aquiring 3 new region's data (including scripts, particles etc), and passing your assets to another CPU. Much of this transfer of data is dependant on network latency.

Even at best, this system is not 100% efficient, because the system is busy with other priorities. Any drop in ping rate, bandwidth, computer performance directly affects the rate in which regions are crossed.

When one sim is in California, and the neighboring sim is in Texas, you are at the mercy of the internet, computer performance as well as a host of other issues.

Simply put, it cannot be fixed.

Modern platforms, like World of Warcraft etc work around this straight line grid issue, by multithreading data relative to the avatar, not the region. In this case, your avatar builds a region around itself and as you move throughout it, so by the time you reach a border, the data has already been passed in advance at a greater distance and more efficient means than the "at that instance" crossings in SL.

So, there's a LOT of data to be moved. Why? Couldn't some of of it be ignored until the crossing is accomplished, or during movement? We have occlusion now, and baked textures, which keep us from having to rezz everything in our vicinity. And I know this has nothing to do with border crossings, but couldn't the same principles--only the stuff that's GOT to be there for now--be applied? Why do I need all the data from the new sim if, for example, I'm just passing through and going to be on the other side in 10 seconds or so?

Well, let me put it in terms of a couple of simple questions: Which of your attachments / prims should go first and start operating before the rest of them get there? Will the events and link messages be able to quickly and easily traverse between the two whilst you are in transit?

For everything to work properly, it has to ALL be in the same context. That means everything has to be transferred at once. Otherwise, you get parts of things thinking and operating like they are in one context, and the other parts thinking and operating like they are in the other context. That's the problem. Simulator contexts are literally islands. They don't know very much at all about what is going on in adjacent regions. In fact, the only thing they know about is what the adjacent region tells them, which is a 10m strip along the shared border to that region. Anything beyond that in the adjacent region doesn't exist. Thus, when you get something that flies into that 10m strip, and then right across the border, there's not a lot of time to start preparations before you simply have to throw the avatar/object data over the wall in a full handoff.

So you want LL to funnel all the business to just 100 creators of their choosinig, and lock out everyone else? Somehow I think you'll be complaining when you're not on that list. Bad, bad idea.

Are there any virtual worlds that use distinct male and female avatar meshes in which clothing items for a female avatar mesh can be applied to a male avatar mesh?

My understanding is that SL male and female avatars use the same mesh.

Novoking had gender specific clothing, as I found out when I bought some male clothes without realizing it but wasn't able to wear them.

Exactly. You hit the nail on the head. Show the noobs only the best builds, the best products and the best non profit regions SL has to offer. Lock out everyone else who is not up to the "top 100" standard. That’s the whole point, show them the best and hide the rest. That slogan should replace that "your world, you imagination" bollocks.

The whole "bad bad idea" as you called it is to capture the interest of the casual SL noob, to present the best first impression possible. Show them the best SL has to offer. It just so happens that a good portion of the top 100 spots in SL are commercial ventures. So what? If the noobs like it and it encourages them to stay then brilliant. The rest of the community will just have to work harder to create a top 100 location in order to benefit.

And you are right, I wouldn't be on that list, but I certainly won’t be complaining, I will be dancing in my virtual street if LL can significantly improve the retention ratio as it will benefit ALL of us including YOU.

Frankly, since you've replied to some of my threads in the past, you're posts are among those I look for. You have the answers to the questions I barely understand enough to ask.

You made a good point about what should go first.

In an earlier post, I said "simplify/compress/expedite" and earlier I used the term "weighting." I can also understand your point about sometimes things have to break before we can move on.

Maybe the way some things work has to change, or things DO have to break before some problems get solved. But losing some of my content, or even some of my work is preferable to starting all over again on another platform because SL closed because everyone decided to go play BM.

Frankly, I'm not sure about Blue Mars yet. That's not a criticism--it's just an observation. Right now, there IS no Blue Mars, and the plug could be pulled tomorrow. I don't expect it to, but it could happen. Google pulled the plug on Lively after only 6 months, and who expected THAT to happen?

To answer your last question, which of my attachments and scripts would I wait for--Turned out to be a good thought experiment. At first I started thinking, "Well my exact position is most important, and then what I look like, what's around me. . ."

But that fell apart when I started wondering how does the grid know the difference between my plane and my hair? Or which one is more important to have in the crossing, so, yeah I see your point.

But at the same time, I know that the grid I started out on in 06 could not handle the numbers that pass for normal today. So, anyone paying attention can see that there HAVE been improvements to the way things work.

As for LL being unwilling to make the changes necessary--Says who, and why not? back in 98 I paid $300 for 8meg of memory for a computer that ran on a 100mhz processor. And those numbers were good for the time.

OK, so I'm way off track in my own thread, which started out on account of my curiosity about the differences in the technology behind Blue Mars and Second Life. And I still am.

Nice theory but many like me who enjoy learning to build here wouldn't like a top100 world, better yet why not just move to the perfect BM world and stop trying to turn SL into a copy of it?

The problem here is one man's meat is another man's poison. Retention takes more than eye candy, people want something to do. Whereas I agree that retention is a serious issue that needs looking at I'm not sure that Linden Lab or anyone else has the ability to come up with a top 100 places out of 27,000 sims or how ever many there are.

Take out the word "copy" and replace it with "competitor". My idea was to offer SL a better chance of competing with BM. Im not trying to turn SL into BM. Im talking about utilising a tiny small fraction of land to better acclimatise noobs for the first few days and then offer some guidance on the best place to visits there after. Im not suggesting that all content must be produced in proffesional 3d external software am I. Im not suggesting that avatars are stuck being the same sex and are restircted to what clothes they wear base don that sex. Im not suggesting that every city has a creator/governer who makes the rules. No I like SL the way it is and im not trying to turn it into BM.

Bottom line is this, if the average noob, and Im not talking hardcore gamers or techies or geeks, Im talking normal people, if the average noob played BM for 3 days (assuming it works as advertised) and then played SL for 3 days (In it's current orientation state) then chances are going by first impressions and shiny graphics then they are going to plump for BM.

This is mostly based on assumtion as I've never played BM but I am believing the hype.

I'm just offering up ideas on how to help SL compete. You dont agree then what do you thinks SL needs to do to compete Tegg Bode?

I think SL is going the right direction, we just need to upgrade our meshes, physics and sound to compete with something like Blue Mars, sure many non -creative people will be swayed by pretty shiney, but others won't whenthey realise the limitations bound on them compared to the freedom here to create.

Says Linden Lab, actually. Many times, many different suggestions have been made about doing away with region boundaries, making "large" regions (512mx512m and up), etc, and LL has steadfastly declined to go anywhere near changing the core of the engine. The reasons they cite are the same ones given here: the amount of effort involved, the amount of hassle testing it and getting it to work, and the amount of existing content that would be irrevocably broken because of it. That doesn't mean they won't consider it sometime in the future, but the past has seen a great deal of reluctance towards even considering it.

Some amount of "pain" is considered "endemic" to the "SL experience". Until that pain is proven to them to be a significant factor in acquiring and retaining customers, they will likely do little to change it, making it better. Not only that, but any changes must mitigate the loss of existing customers as much as possible.

When you have content which is actively being supported and developed, upgrades and changes to the underlying platform which break that content won't be too painful, because the developer of that content will adapt to the changes and send out updates to his/her customers. However, there is a large body of content which is more or less "abandoned". The developers are no longer in SL, or no longer interested in developing/maintaining content in SL, for whatever reason. When that content breaks, it's done; ready to be tossed into the bit bucket. That is the content which, when permanently broken, will be a significant factor in a resident's desire to stay or leave. To them, they paid good money for a product, now the world context has eliminated that product's usefulness or aesthetic value to them. Essentially, obliterating their "investment". Thus, it isn't something that most people will look upon favorably.

I disagree. I am a creative person. I have a successful business and a very large stake in SL, however, I am often frustrated with the limitations of the amateurish in-game building tools. I have been dabbling in 3ds Max for years and I am very excited about the possibility of being able to create, for example, a house in the full freedom of a real 3d program and then bring it into BM and walk around in it. Blue Mars isn't putting any restrictions on who can be a content creator or on what you can create. The registration system allows them to protect a content creator's work from being stolen.

Fact is, it's going to be a while before we see exactly how much avatar freedom there will be in Blue Mars. It's all still too beta to know. I suspect that SL had a few restrictions at this point in it's development too.

That hasn't been the case for years, except for out-dated hardware that only supports fixed-function hardware T&L. Any game with a relatively modern shader-based rendering pipeline can do per-pixel lighting that is near perfect with a quad of any size.

.

How outdated? gf7000 series? gf6000 series? Radeon 8500/9200?

At least as far back as gf5700 I've been using SM 2.0 since 2003, so let's call it six years...


.

And how about Intel Extremely Painful GMA9xx chipsets?

Dunno. What about them? Are those considered a modern video card supported by most modern games? I'm guessing by the fact that you brought them up as an edge case that they are pretty much crap, so I don't see how that's relevant to modern rendering technology.

If you have a video card that doesn't support a reasonable implementation of shader-based rendering, you simply have to accept the fact that it can't do everything you might like.

.

I wouldn't recommend trying to run SL on an Intel GPU myself, but they're what you get in a typical laptop or business desktop. Many people in SL don't play games. At all. They don't think of SL as a game. They might, after some time in SL, decide to upgrade to a better computer for SL... but not if it can't run on the computer they already have.

That's true, but it also serves to prove my point that the SL rendering pipeline is horribly inefficient. Actually, never having had to use a card like that, I don't even know if SL will fall back to vertex lighting. Won't it just disable local lighting altogether?

Second Life does try to cater to the minimum-spec machines of the world, but there are some things even SL won't try to compensate for. It's an unavoidable fact that not every machine is suitable for every task, and if that excludes or alienates some people, well... That's life.

.

Uhh.. so you're saying new hardware doesn't support vertex lighting anymore? O.o

Per-pixel lighting is expensive compared to vertex lighting, which means that you can't have as many of the former versus the latter. Also, vertex lighting is better for global illumination, such as from the sun, or other bright infinite sources.

Regardless, SL uses vertex lighting only right now for point lights, so while it may not be the case elsewhere, that's what we currently have to work with.

Maybe they will get around to adding shader-based per-pixel point lights eventually. That will be great when they do. However, the economy of vertex lights still will be useful, and can justify keeping some level of extra tessellation.