Lets be honest

[Dogboat Taurog]

 

---

Agent Tairov wrote:

 

---

Dogboat Taurog wrote:

 

i'm gonna be honest, i don't believe you. and quite likely nor do most of SL.

you want someone to argue with, thats all.

---

 

Actually, I don't. On top of that, you're either a troll or deliberately inflammatory, and apparently you don't care for actually backing up your arguments. I don't have time for someone like that. Bye.

---

 

not at all, i have tried kirstens, LL V2, LL V2 beta and firestorm i find them all slower than V1.

what backup do you want? you come in here calling troll without anything to back you up.

maybe all the V1 users are trolls?

maybe you are the troll...

[Cerise1488303085]

For the last few days I have been taking mesh development viewers out on the main grid, on low end modern hardware. The difference in performance is positively stunning, to the good, even with only the main grid's pre-mesh content to see. It is so much brtter that I don't even think mesh will be the big draw for these new viewers, they put the older and present production versions to shame.

[Dogboat Taurog]

mesh on the main grid?

what viewers are you using and where?

i would be interested to see this.

are there non V2 mesh viewers?

[Gaia Clary]

I guess, she uses v2 development viewer on the main grid.There is nothing wrong with that. And maybe that is also the cause of all missunderstanding ? I haven't tried any other viewers besides the mesh development viewers and Kirsten's viewer for a long time. They just look way better to my taste (regarding graphic quality), They seem to be more stable (i can not remember of any crash since many weeks) and maybe they are optimized to better use the capabilities of the (my?) graphic card ?

What i have seen recently was, that these viewers also work almost twice as fast (for me) after i migrated from 32 bit windows-7 to 64-bit windows-7. Nobody could tell me why that performance increase happened just from upgrading the operating system, but it was very very apparent and i was delighted to see 60-80 frames per second where i could only see half of that before ;-)

[Dogboat Taurog]

 

---

Gaia Clary wrote:

I guess, she uses v2 development viewer on the main grid.There is nothing wrong with that. And maybe that is also the cause of all missunderstanding ? I haven't tried any other viewers besides the mesh development viewers and Kirsten's viewer for a long time. They just look way better to my taste (regarding graphic quality), They seem to be more stable (i can not remember of any crash since many weeks) and maybe they are optimized to better use the capabilities of the (my?) graphic card ?

What i have seen recently was, that these viewers also work almost twice as fast (for me) after i migrated from 32 bit windows-7 to 64-bit windows-7. Nobody could tell me why that performance increase happened just from upgrading the operating system, but it was very very apparent and i was delighted to see 60-80 frames per second where i could only see half of that before ;-)

---

you must have a beast of a PC.

 

i'd still like a definative answer though.

[Cerise1488303085]

 

---

Dogboat Taurog wrote:

mesh on the main grid?

what viewers are you using and where?

i would be interested to see this.

are there non V2 mesh viewers?

---

 

They are here. The downloads in the brown box have the most up to date tweaks, but they are experimental builds so some could work and others not. Right now the status page says "build results pending" but after that updates, you can try it.  If anything is wrong with today's build, there will be links on that page to the last good one, which does work OK for me at least.

It will be good to use a separate cache folder for experimental viewers, they already use independent settings files. This viewer is really made for the preview grid, so it defaults to Aditi logins. Remember to change that to Agni.

I do not think there will be non-V2 mesh viewers, the 1.x code is too old and slow to adapt it well.

[BlackSoulAngelFire Viper]

oh and, i forgot to mention about the reason for the post,

 

 

http://funpresident.com/2009/06/starcraft-karrigan-breeding-of-a-sculpture/

 

i want that avatar, but, guess what?

 

ive been to a fair few great creators of things in sl, including the creator of the stuff in violet studios, which has a VERY wide range of VERY good quality stuff

 

guess what they all say?

 

no way, not till mesh is out, due to the very fact that trying to make that, out of sculpties, would just end up being a mockery of it.

 

 

mesh is needed, or else sl is going to fall behind, and just like horses, who used to be the main form of transport in the world, it will be left behind, something to look at, and laugh to ourselves as we wonder how people could have ever seriously used them, compared to the now up to date things we have.

[Tegg Bode]

I see sculpties being a bandaid on the way to mesh. As far as I understand  sculpties are a limet form of mesh. Mesh is much more powerful and more content available, I don't see any game engines upgrading from mesh to sculpties.

Sure there are going to be issues etc but that probably why they haven't been released yet.

The biggest problem I see for the consumer is shape & sizing of bodies & clothing, hopefully some sort of standard will occur. I'm hoping most items will still stretch and shrink to enable some fitting to be possible for bodies that are taller or shorter and slightly different proportions.

We need to move into the 21st cntury sooner or later, despite a lot of good efforts our 90's avatars suck.

[Gaia Clary]

---

Tegg Bode wrote:

I see sculpties being a bandaid on the way to mesh. As far as I understand  sculpties are a limet form of mesh. Mesh is much more powerful and more content available, I don't see any game engines upgrading from mesh to sculpties.

---

I think it is important to understand why sculpties where introduced at all. And why they have been introduced in that particular way.

Before we had sculpties, there where prims. And prims where unbreakable blocks of building units. We had 15000 prims available on each SIM. And the entire infrastructure for these building blocks was optimized for an "online game engine with dynamic content"...

But Prims came with limitations regarding modelling capabilities. It was amazing to which level you could get with prims only. But time moved on, and the desire for even more expression came up. So the next level was ... meshes.But now there was a problem (or two...):

Since the entire infrastructure was setup on having primitive building blocks, (arbitrary) meshes would not at all fit into that scenario. How to evaluate a mesh ? How to compare its "value" with the value of Prims ? How to keep it stupid simple for the users ? ... And how to keep the existing infrastructure at all with meshes ?

So it was necessary to created a sort of "extended prim" which stll was sort of a primitive building block, but had more capabilities regarding modelling. It must fit into the existing system, should not break anything and it must be efficient like nothing (that was the goal at least).

The finally introduced limitations where:

 

• Use a fixed mesh topology: The mother of all sculpties is a rectangular surface bended and stretched in 3 Dimensions
• Use a fixed texturing scheme (UV-map): All textures work the same way. The system always knows how to wrap a texture around a sculptie. Nothing to take care, all the same, no extra data to be downloaded for displaying sculpties (no uv-maps, no normal maps)
• Use a fixed amount of vertex space per building block: to fit well into the existing infrastructure of Prims.
• Use the existing data-containers: Iimages where the only binary content allowed for upload to SL

And the final result where Sculpted Prims with all their Quirks. Please understand that Sculpted Prims are an invention made by Linden Labs (or for Linden Labs) to fit into their infrastructure! So you will not find Sculpted Prims anywhere else in the world, except in SL-clones...

Now we are (hopefully) closer to regular meshes than ever. But the original issues remain: How to evaluate, compare, integrate meshes into the existing infrastructure without breaking the whole existing content ?

I can imagine that solving this equation is a nightmare. But is there an alternative ? Creating "Third Life" maybe and just forget about SL ? Wasn't it what Blue Mars tries to provide ? A better environment in all aspects ? Did they succeed ? I haven't looked into BM for a year now. Maybe i should do to see how they evolved...

And to get back to the original question in this thread: I doubt that "enhancing the sculpty features" is the correct move. It makes much more sense to me to enhance the infrastructure for building blocks in SL instead of optimizing the building blocks for the existing infrastructure. And that is a tedious and time consuming process (as we can see). But i am sure that once meshes gets to main grid, that will boost creativity, usability and experience.

 

---

---

Tegg Bode wrote:

The biggest problem I see for the consumer is shape & sizing of bodies & clothing, hopefully some sort of standard will occur. I'm hoping most items will still stretch and shrink to enable some fitting to be possible for bodies that are taller or shorter and slightly different proportions.

---

The resizing issues with avatar clothing are only related to "rigged clothing", aka clothes which move along with the avatar skeleton. If you just attach non rigged meshes to the avatar, they can be scaled like any other building block (prim,sculptie). So when you are not interested in rigged clothes, then use non rigged clothes and everything is good for you ;-)

[Nathan Hopkins]

I think Dogboat is expressing a legit concern.  I have no idea if mesh will create performance issues, but if it does, LL cannot afford an "upgrade your machine" mentality.  A lot of people are worried SL's slow decline in use won't turn around, and so LL ought to try to balance upgrades and accessability.  I don't think pointing that out is unreasonable.

The worry about needing to use V2 also seems legit to me.  I've been in SL since the middle of '06, but actually switched to V2 when it came out.  I found the experience of learning a (what seemed at first) slick new interface sort of fun, and liked some of the new features possible on V2 that were slow to be included on Phoenix.   AND, my main viewer now is Firestorm.  But despite that, I still literally el oh el while using V2 sometimes.  It really seems like it was made by people who had never used SL but were given a thorough description of what it is.  As much as I think the V2 bashing is excessive, what LL's customers think is all thats relevant, and in that sense V2 was a major disaster for LL (again, I don't mind it and its what I use now).  So if V2 is required for mesh, and mesh objects start to fill SL, there will be some lost customers over that issue... a % won't use V2.

Also it seems one of the few reasons LL doesn't have a real competitor is the access people have as participants in its economy.  Mesh might end up being a bit of a barrier to that, which also may effect SL's appeal for many people.  I don't think that is reason enough to not do it, but these down sides taken together ought to at least be thought about by people invested in the platform and who want to see it succeed.  The idea that "LL knows that its doing" shouldn't put anyone at ease.

 

[Drongle McMahon]

There is no doubt that one of the main motivations for introducing mesh is to reduce the load on the gpu by reducing the numbers of triangles needed for a given shape. Sculpties and all but the simplest prims are very poor for rendering efficiency. Meshes can be much better. However, it is alo easy to make inefficient meshes with excess triangles. So a major effort of the developers is developing a costing scheme that will make inefficient meshes impracticable. Making the same thing from meshes and sculpties will generally cost at least as many prims as the sculpties, and sometimes a lot more, unfortunately. This is a complicated and difficult balance to achieve.

The trade-off for the increased rendering efficiency is the bandwidth requirement for downloading the mesh data. This is more or less the same sort of problem as texture download, but once again mesh are subject to a costing scheme that genuinely reflects the download resource consumption. This should ensure much better control of excessive mesh data use than of excessive texture use (i.e. no control whatsoever in the case of textures!).

Unfortunately, none of the cost-related limitations presently apply to attachements. As things stand, mesh attachments may be as bad as sculpty attachments for excessive use and lag for those with slower connections or old machines. Perhaps a solution will be forthcoming.

One problem that remains is that it is easy to produce high resolution custom baked textures for meshes that greatly improve their appearance. They can be made for sculpties and prims too, but it's a bit harder. These cannot be reused on different meshes. This may substantially increase the use of large baked textures, with consequent problems for texture download times as long as there are no texture resource use limitations. It is quite possible to make UV maps for meshes specifically to enable the use of small re-useable tiled textures, but this does take a lot of extra work. As well as the texture download savings, this has the advantages that anyone can re-texture the meshes as they wish. Let's hope this practice is adpted as widely as possible.

 

[DanielRavenNest Noe]

 

---

Gaia Clary wrote:

---

I can imagine that solving this equation is a nightmare. But is there an alternative ? Creating "Third Life" maybe and just forget about SL ? Wasn't it what Blue Mars tries to provide ? A better environment in all aspects ? Did they succeed ? I haven't looked into BM for a year now. Maybe i should do to see how they evolved..

---

 

Blue Mars for PC development is on hold at the moment, because they ran low on money and let go 2/3 of their staff.  Blue Mars for Mobile devices is being worked on by the staff they have left.  The PC version is still online, but it is in an incomplete state.  I am painfully aware of this, since I own two cities/game levels/sims there (they are larger than SL sims, so its not a direct comparison).  The plus side is I got to learn how to make 3D models (ie mesh) - everything on Mars is made of them, and I am selling a decent number of clothing items, mostly on the iPhone/iPad version of the Mobile client.

Blue Mars uses the "CryEngine 2" graphics engine.  That's the part that processes your models and textures and turns it into a scene on your monitor.  To make a virtual world, you need to add chat, friends list, a shopping system, a way to customize and dress your avatar, etc.  The latter part is what was not finished.  The engine was licensed from the company that made it (CryTek) and worked fine right from the start.

There are a couple of other general purpose graphics engines available, such as Unity and Unreal, and some efforts are underway to make virtual worlds using them.  All that other stuff besides how it looks, though is more important to the end user. 

[Medhue Simoni]

Sorry for progressing this off topic conversation, but I could not resist.

I had high hopes for BM, considering they were using CryEngine. The platform is known for it's environments. The first time I entered BM tho, yeah the environment was there, but I could not understand why all the other elements of the CryEngine platform was redone. Years ago, I spent a good year or more creating MODs for the original CryEngine. The Editor for it was down right amazing. Of course, I have no understanding of how it all works, but seeing BM for the first time was a bit of a let down, considering my past experiences with the engine.

For me tho, the bottom line was the animation system. It looks like the BM team spent massive amounts of time on their system, but it was not made with the merchant in mind, or in anyway flexible. LL's animation system is way beyond it's time, and is only now showing it's vast superiority. Plus, BM really messed up by not making the system to fit bvh files. BVH is the most common motion capture format. To make a system that does not allow for the importing of bvh files is like making a sound system that doesn't play mp3 files. Oh, lol, like SL's sound importing, but we can still stream mp3's in.

[WADE1 Jya]

QUOTE: ortiga Waco "By doubling the amount of vertices allowed in a sculpt Linden Labs would stress their servers less than with mesh, and keep a building community thriving."

All you end up with then is a hard to work with mesh with lots of annoying built-in limitations?

No thanks, I'd rather just have the total freedom of normal meshes :matte-motes-inlove:

[Nyx Linden]

I haven't had the time to read through every post, but I've been keeping track of some of the ongoing concerns. I'll try to answer a couple of them here:

1) Meshes are no more difficult to render on a per-triangle basis than normal prims or sculpts. I've done some basic performance testing, and given avatars of roughly equal geometry (by triangle count), there is no noticable performance difference. There is a slight performance hit with rigged meshes but it fell within a couple of percent.

2) Meshes should be able to be significantly more expressive on a per-triangle basis than previously available content (prims and sculpts). The equivalent in geometry of the fully detailed mesh avatar I was using was 5 prims. Even the most efficient avatars I've seen use significantly more than 5 prims.

3) Making sure that meshes don't cause performance concerns is mainly a function of making sure that our algorithms for how many "prims" a mesh is equivalent to is correctly tuned. These functions are not finalized yet, and we plan on working with a lot of test content to try to get these numbers right, to get the right balance between offering incentive to use mesh and ensuring that we're not increasing the geometric complexity of the world.

4) Doubling the number of vertices in a sculpt would not only increase this geometric count (which we're trying to not do with mesh), but would not provide anywhere near the power and control that allowing mesh import would allow. In addition, sculpts have fixed (poor) LOD management and loading behavior. We've made great progress on the mesh import project and believe that it should allow users to create much more expressive and much more efficient content than extending the functionality of sculpties. Thanks for the feature suggestion! We're going to focus on mesh import for now though.