3rd party imports

I've done some calculations in my Windows calculator and the calculator says "No".

Well, it wouldn't be the prim system anymore... Remember, vertices aren't fixed with the way SL generates stuff, it depends on how detailed the model is being displayed on your system at the time.

Unless you want to manually do each possible level of detail.

<Logan runs off crying>

The thing is, though, that:
- It's important that people should be able to create silly things in the sandbox.
- It's important that people should feel that, by creating those silly things in the sandbox, they are using (and in fact, learning) the same tools that produce all the cool stuff they see around the SL world.

There is not a single newbie I have ever met who has not reacted with a "wow" or "whoa" when they learn that second point. (Except for those who arrive already knowing it of course.)

And that second point is what's often responsible for these people trying to become good builders. If you say "well, you can use this for silly stuff but for the really good builds you have to use an external 3D modelling tool" then that sense just isn't there, and people won't feel motivated to try to become good builders unless they were 3D modellers in the first place.

Kinda like photoshop and texturing

Oh look I made a short comment for once.

Importable models would allow for some pretty amazing opportunities for about 0.5% of the user base. It would also allow for a bazillion identical uploads of the exact same Star Wars space ship mesh. So while we'd see some pretty cool stuff from a few people, we'd see the exact same things from everyone else.

I don't think that's an argument against custom mesh importing. I think once the technology is feasible, it should and will be implemented. And, fantastic, I'll have yet another thing to teach myself.

I am not a Blender user myself, so I leave that to any Blender fanatics to argue.


This and previous posts are equally good arguments for ripping out the ability for us to upload textures. LL can just give us a really dumbed down version of PS in game and we can create textures socially. Lets reinvent the wheel twice. Afterall if someone is truely talented, they can make it work. I am sorry that I don't get the "lets make things harder" approach. I can see having a basic version of the tools to place objects and to get people started in 3D design. However, I wonder if all the development time they spent on the build tools they could have spent on other things.


Right now by depending on limits on prims, builders have been able (inadvertently) knock client performance into the dirt. I can make a 200 prim object that is (hopefully) under 3k triangles or it could be made up of 300k+ triangles. Yuck! SL is always going to have client performance problems until they deal with that. If you were to talk about tri counts to buiders, how many would know what you were talking about. It's not completely their fault as it is it can be tricky getting an accurate tri count on an object.

I think builders have been hitting the limit on the build tools for quite a while. The trick is finding a good builder who is also a good texturer in one package.

Regardless, this whole poll is pointless. According to other posts, we aren't getting such fuctionality even in SL2. So, the soonest we could see it would be (generously) several years down the road. If we get it at all. We are more likely to see the ability to upload meshes in a competitor's product.

Sure. I'd be all for, at some point, adding a version of GIMP or something to the SL client. At this point, though, we already have a working system . Just like we have a working building system. Both are, at this point, re-inventing the wheel to please a few people.

But I'm not saying /make things harder/. I already refuted that once.

The point of SL was to make content createable for anyone. That's the whole "your world" thing. They /had/ to make the 3D content buildable in world for that to work, because, as I said, good 3D packages are expensive and difficult to learn. Good 2D graphics programs are much less expensive and easier to learn (Even Photoshop doesn't compare to most 3D graphics programs).




I'm not arguing that the SL rendering engine needs work, or that it's building engine might need work. But adding 3D importing wont solve that, unless you also take away the ability for other people to use the in-world tools. People who know enough to care about such things already do in SL, and it's controlable to an extent. People who don't care aren't going to when they are importing models, either.

I think builders have been hitting the limit on the build tools for quite a while. The trick is finding a good builder who is also a good texturer in one package.[/QUOTE]

Nonsense. I agree textures go along ways here, but have we hit the limit of what prims can do? I don't think so, not by a long shot. And if LL listens to some of the suggestions for tweaks to the existing system, such as offset hollow and such, well...

Well, I know you said it in jest, but it is true. Of the new folks I've spoken to, many want to learn to build. Some want to learn to script. Very few want to learn to texture.

I remember a program that I started with called "Simply3D"
That was about as simple as you can get. I picked it up as a kid.
And it is far more complex than the tools we have now in SL. But very simple to use.

It featured large vector points to move around in order to shape your object. Building tools can't get much simpler than that program. The learning curb of point and drag a box (vector points) isn't to hard.
That program also had certain shapes to start with, just like Prims in SL, that couldn't be changed other than that vector editing.

The amount of data to send for each prim is the location, all the twists and what not settings, Creator data, etc.
Seems to be no more complex than a bunch of vector coordinate unless it was an object of 10 or more.
Even if we had a max of 10 vertices to an object. That would allow custom morphing of that object which is not available right now.
I don't see how the data could be higher that what prims are already. Because it's not like each vertex needs twist data to be sent.
If so then what is this other data that is needed which makes it larger than the current prim data?
A point in space needs 3 values xyz.
3 x 8 vectors to form a simple mesh cube = 24 values.
OH ehah. Nevermind.

Well - here we go again. I'll park the "yay I made script" bit at the door this time, since it's not really the point.


I like to slice this issue into three parts, all of which I think are quite relevant:

- Technical Feasibility
- Impact on Creativity
- Overall Utility



For that first one, from a technical standpoint, Second Life's system is built to utilize prims - generally fixed length streams of data that, while good in concept, are only half-baked in execution. Prims are great for rapid prototyping, but they're still a longshot from being both efficient (ie. face culling) or useful on the same level as professional manufacturing or game CAD tools.

That said, prims are still a good compromise. They are not, however, a final solution.


Using that as excuse to move on to creativity, here's why. Prims, while good for general "rigid" modelling, are horrendous where "soft" and "compact" models are concerned. A billboard face, for example, is six sides - as opposed to two or even just one in standard modelling. This adds up. Similarly, prims cannot be "rigged" in the same way of traditional meshes or NURBs, creating a problem at joints where the intersection has a visible break.

Similarly, prims - by their nature separate entities - create lighting situations that can be jarring at best. While the argument can be made "just blend them together!" - that would take a fair deal of wasted processing power to do on the fly.

What I'm getting at is this - while prims are just dandy for more "charicatured" or simplified design, they cannot hold a candle to more photorealistic or full production modelling. Many designers, to their credit, have come pretty close with rigid-body objects - but they're fighting an uphill battle against a system not (yet) built for their needs.



As for the overal utility, I see it less as "what we can do with prims," and more "the world still uses different systems." Adam raised a good point that mesh data can be quite compact - my numbers for Blender models are more or less in line with his assessment (4-5k of polygons in ~200k).

But the real utility, in my opinion, is getting the rest of the world interested in building content here. They won't do that with a proprietary system built a grade lower than what they're used to. But if it's something familiar to them, something they already know and use, crossover would be minimal and cause much greater interest in what is (still) just a niche product.



I've tried getting a few of the Blender community interested in working with Second Life, including a Summer of Code entry. Their response was unerringly: "Why?" My Summer of Code entry was quickly thrown out.

While prims are a neat compromise, and Enabran is correct that most existing SL users wouldn't have a clue, it would be a major step in taking Second Life to the creative mainstream. But, since the system is not built that way and probably will remain so, I think it's a lost cause here at this point.

My fear is, however, that in making SL appeal to one group - the small amount of professional or semi - professional modelers out there who refuse to use SL soley because of its existing system (Lets be honest, the modeling is only one facet of SL... If all you want to do is model, SL will *never* hold a candle to doing it the 'old fashioned' way) - you are going to end up *alienating* a much larger base of people who will now be second-class creators compared to the pros, but not by virtue of skill or talent alone, which many people in Second Life have in abundance.

And that basicly sums up my fear, I guess...

(People who wouldn't want to live in a world of "look but don't do" next to the pros) > (Pros who won't 'lower' themselves to using lesser tools, but would otherwise be SL's target audience.)

With all due respect, I think you're marginalizing a group that is actually much larger than "the small amount of professional or semi-professional modelers out there who..." This has not been my experience working with people of more simple tools like Anim8or, many of whom are just random people interested in giving it their best shot. I'm not exactly a pro, either. Anim8or is where I started out.


The nasty part is this. If prim building is at the clear disadvantage, the better system will prevail even if it doesn't happen in Second Life. It's one of those "necessary evolutions" I've worried about since my early days here, which coincidentally, is why I don't have a lot of models stacked up.

And I'm working on a conversion of prims to mesh for this reason (evade doom!).

I think this debate about mesh download sizes misses an important point.

Client lag is usually described as a function of the amount of polygons the client has to draw in a given scene at some resolution. Even if you could download a mesh file quicker than a jpeg, the presence of a lot of meshes in the environment would lag everyone to a standstill.

Some Maya afficianado with a super quad-Gig whatever PC would make a store that lagged like a room full of hoochie girls and never even realise that the rest of us couldn't walk through the door.

Small compared to the user-base of people who don't meet those requirements, I will absolutely stand behind. Without reservation. Note that I said, "who will not lower themselves to use other tools". I meet the definition of random hobbiest with 3D software, and have done modeling for other games such as Quake (up to 3 inclusive, not including 4), Jedi Knight, Total Annihilation, and a few others. But I'm willing to use what SL gives me... I don't hold myself 'above' the system.




Maybe. I'm not convinced it's a "clear disadvantage", though, not at this stage in technical progression.

This is a myth, sorry. But it does touch on another important point.


Prims are the way they are now to simplify the concept of faces into coherent "objects." Having a sim limit of 10 or 15k of prims is a lot easier, both at the database level as in the viewer. Prims are basically fixed-length "records" in the database. JPEG2000 of powers of 2 are used, I suspect, for the same reason. Pretty smart idea on the Lindens' part.

In contrast, raw polygons would run the course anywhere from far more efficient than prims to standstill lag, as you mentioned. But this gets into another assumption of Second Life not likely to change: the "one shared world."


As it stands, the Second Life system is an amalgamation of compromises that aren't likely to change because they're fit so tightly together. The only way these assumptions will change is with a completely different product or radical reworking to the point of being a new one anyway.

Getting meshes and then pro modellers into Second Life is what will help to make the place look nice. The most popular multiplayer games are "look but don't do". Second Life is already alienating the large majority of the world's game playing population. But meshes alone aren't going to save the world. There is still the problem of streaming 3D which will never be able to meet demand. The faster our connections become the more detailed the models will be. We'll always be waiting for the 3D to download.

Then what's the point of Second Life? Might as well just go play World of Warcraft.

Second Life exists, and continues to exist, because that's not what all people want.

It's really not about "too good to use Second Life's system." It's Second Life trying to strike out with a proprietary system the rest of the world just does not use, and with good reason.


Second Life models are not useful anywhere but... Second Life, with little or no way to back them up. That does not inspire confidence. Further, tools from the examples you gave, including QuArK, Worldcraft, Hammer, and Torque... are. A model in Half-life could more or less be converted to JK2 with QuArK, for example. Another good study in this is Milkshape, which does much the same.


I'm not saying you're wrong to like the primitive system - it's a great tool! But for what it's worth, it's too closely held and, well... "primitive" to make the industry want to even bother with it. I'd be willing to bet, given every Second Life developer, there are three people that would be interested in trying it if it just used their stuff.

And that, in my opinion, is a big thing.

Very true, Reitsuki!

But Second Life is appealing to minorities. Builders are a minority. I think most of us would like to see it appeal to a majority.

Wouldn't we?

I would disagree, since spending many hours on Help Island since its inception, the biggest thing people want to know is.... "How do I build!"

GrrrrRRrr.

Yes, that is the people who join Second Life to build. Of course the ones who join to build are going to ask you "How do I build?". I'm talking about the multiplayer games playing people in general. The people who just want to have fun. All those hundreds of thousands of people who are playing WoW right now.

Jeffrey summed up a lot I was going to say anyways.

But if SL either pulled developers off another project like Mono or Havoc2 and put them to work or hired additional developers to work on making a stripped down version of GIMP or PS inside of SL, someone at LL needs to be slapped. What will we end up with? An outdated version of GIMP/PS as the LL developers can't keep pace with those two products as they don't have the resources. And whatever new features we would have gotten in SL get their schedules pushed back further and further. Great.

While I am a 3DSMax user, here is a list of free 3D modeling tools. I have a hard time believing that they all suck: http://www.3dlinks.com/links.cfm?categoryid=1&subcategoryid=2. We don't need photorealistic models, we just need something good enough for SL. But I don't know about you, but I have met way more excellent builders than I have met texture artists. The good building texture artists in SL are few and far between.

Interesting, cause I have gotten that from so many other supposedly more knowledgeable people than myself.

I am not disagreeing, I just find it interesting how so much of what we think we know about SL is really just hearsay and conjecture. Another argument for some kind of documentation of SL itself and how it works I guess.

So if it's not the polygon drawing that slows down the CPU is it the calculation of the tori etc. themselves?

I always assumed that since the actual data for the prims is minimal that it was the clients translation of that data into polygons that determined a great deal of the client side lag. ie - more comlicated shapes like tori and avatars cause more lag because they take more time for the client to draw.

Custom 3D meshes are quite nice to consider in the "LONG run". The SL code just doesn't feel at its optimal/stable point now and I feel this needs more attention for the time being. I could see a big market (though very limited at first) for custom meshes. Creatures, mutations, thingamajigs and whatnot. Also let's not forget with custom meshes comes custom skeletal rigging and UV mapping as well. The shape has to have some internal structure and accurate placement of texture ya know. Imagine 1,3,4,5,6 legged life-like figures. Tails, limbs, etc with their own custom animations based on your skeletal rigging. Not the most attractive to general audience (EDIT: creation wise). Mostly for the professionals or those really willing to try for it

Yes, it is frustrating dealing with the upsets of SL. Yes, there is certain limits in place we have to work around or deal with. I personally think the true creativity comes from knowing how to make the most of what you have now as silly as that may sound to some folks.

Eventually though the tools will become more receptive to possibilities/capabilities and without much "give up in frustration" due to particular limits/issues/etc.