Are we overdue for a new AV mesh?

I've been looking into Zbrush, recently, having mastered Maya and wanting to explore something a bit more kooky. (both are 3d modelling programs)
A friend of mine at Industrial Light and Magic said he'd show me around ZBrush - since he'd been playing with it too - and we would look into how to import stuff into Second Life through its pipeline. We found a link to download the current avatar meshes from second life and uploaded it into Zbrush.
As soon as the mesh appeared his first words were. 'This is undoubtedly the worst model I have ever seen. I dunno where they got it, but this, frankly, is terrible.'
'What's wrong with it?' I asked. He just shrugged and said 'What's right with it would be a shorter answer.' Apparently it looked about 7-8 years out of date as mesh technology today goes and suffers from multiple problems the main problem being that it lacks polygons in the important places (IE, the crotch, the breasts, the armpits and shoulders).

This prompted me to think. Are we long overdue for an avatar overhaul?


Bear in mind my friend is a professional in his field such as I am a professional in mine (he does 3d, i do 2d and a bit of 3d). So, if someone from Industrial Light and Magic can say 'This thing is the worst mesh ever' does it mean we need a new one? Pretty quickly? Since Second Life is all about being commercial these days wouldn't it be in their better interest to concentrate now on making a new Avatar Mesh?

I might just add that it was mentioned to me that it was 'difficult' to replace the mesh.
I'll put my two cents in to linden lab to say that it took us only a few hours to update and restore their mesh. They can even have the project file for free, if they want.

One concern is that changing the avatar mesh would break content: you'd pretty much have to have new asset types for body shape and all clothing assets, and you'd have to have both meshes in use concurrently for a very long time so people can still use their old clothes and skins...

So, a new avatar mesh still sounds like a good thing, as it would necessitate a whole lot of new content generation/purchases! Ask Cisco if forklift upgrades have ever stimulated profits *laughs*

A serious question: of course animators would want to create new anims to take advantage of a new mesh - but does anyone know if it would break existing anims, or would they just look needlessly clunky?
.

Having "Avatar 2.0" available to us, with higher poly counts, better UV mapping, both arms on the UV template instead of just one arm... has been a highly-requested feature upgrade since before I came to SL, over three years ago.

The big issue is one of compatability. Say you release a new avatar mesh, with 10X as many polygons, two arms fully mapped, and much better UV mapping...

None of the existing clothing or skins would work on that new avatar.

None of the clothing and skins made for that new avatar would work on existing avatars.

It's highly unlikely that the existing avatar animations would work the same on an avatar with a new mesh, unless the rigging of the joints was exactly the same.

Everyone's Client software would have to be upgraded to properly display the new avatar, and to somehow distinguish between the old avatars and the new ones. Anyone running the old client would be unable to properly render the new avatars at all.

When Poser or Daz Studio comes up with a new character mesh, it means that the artists who want to use it have to buy all new clothes and accessories to work with that new mesh. But the artist only has to worry about having one to a small number of characters in any given render, and they choose who shows up. It's much more complex when you only control one one-millionth of the avatars you might encounter.

Although I think the current avatar's mesh is very FUBAR, especially the male one, I seriously doubt that we are going to see a new model REPLACING the current one anytime soon. Two reasons. One: More polygons are meaning more load are meaning more lag. Two: A new mesh would render huge chunk of inworld investments useless and reset (if not break) a huge part of the already struggling economy. Both would lead to problems which LL could not cope with easily.

Thanks for explaining that Ceera.

The JIRA issue:

http://jira.secondlife.com/browse/VWR-3432

omg agreed. ;0 did you start your friend at ilm around the ankles?? ;0

p.s. oh, you said what areas you already looked at. ;0

p.p.s. if there was a way to do an upgrade while remaining compatibility across a certain period, via mirror-copying the current user db, i'd say 'bring it on!' ;0

but that's unlikely without a lot of heartache... i'd say realistically we'd hafta be given a 'one-year's heads-up', with working final example matches, to work with, to be able to do so.

actually, i'd categorize this as 'inevitable', because... well, it is, either here or somewhere else. lol don't even -start- me on the crotch... ;0

(actually an old friend of mine has a joke for that... 'what is the name of that stretch of skin between the ____ and the ____?' well the answer is, 'the perineum' ;0)

p.p.p.s., 'as goes porn, so goes tech,' as the saying would go anymore.

Nods.

Sooner started, sooner ended!

One of the plus points would be that in time, we could thrash most of all the " freebies", full perms BIAB and stolen content relating to AV products. I would be curious to see if we had learnt anything from the first time around.

:0 that being said, there's been long-rumoured that a ll/ibm group was working on 'updated intercompatible' avatars, and i presume that includes meshes. but i've no insider scuttlebutt to relate of that anything more than the initial reports... hopefully at least that's why ibm's keeping a campus inworld. ;0 (i don't think i've heard of them pulling out, at least...)

ask your professional friend from Industrial Light and Magic how long it takes to render one of his avatars.
and then make one that does not break everything in SL made for avatars that so much money has been spent on.

:0 there's big talk i think around nvidia (or some other g-card manu, but that's right isn't it?) about real-time ray-trace?? anyone have stats for that? gosh sounds interesting, esp. related to true-mirror/shadow rendering...

Is it possilbe? To make a new mesh that would allow us to continue using our textures and animations while at the same time allow a new set of textures and better animation? Or is a complete reboot of everything necessary? I mean if it's impossible then we can stop wishing for it but if it is then why hasn't it come yet?

We don't need more polygons really, but we do need a better management of them, many zones of the avatar have a shitload of polygons when they don't really need that manyto achieve the same effect.

Im new but Ill put my two cents worth in. I think they should make the AV mesh with a whole torso so things match better. There are some things different side to side with them that I have noticed and its annoying to me when Im trying to get things even. I didnt realize they were that old. Now that I do I can see that there are many things that could be better but, alas, I cant work in 3d very well yet so who am I to be a critic. lol

My sentimate exactly... while the friend at ILM would have the luxury of virtually unlimited bandwidth backbone to render his ideal Avitar mesh, I am sure that any considerable alteration to the existing mesh would cause a serious bottleneck to the already fragile nature of the grid.

While something new would be exciting.. think of what would also need to be replaced.... Dances, sits, yes even the sex stuff too.. I am not an expert on animations, however the way an avitar is animated is on a bend point or axis if I am correct (i could be wrong too..) therefore you are looking at a whole slew of items made and purchased that would be rendered useless... just my opinion

This is something that started out flawed, and after so long, it's horribly difficult to change without annoying most of the userbase. I wouldn't be against having to start from scratch again, but then I don't have a huge wardrobe, and I don't make clothing or animations.

By the way... a new mesh doesn't mean it's going to be super high-detail. Some people don't seem to get it. It can use the same amount of polygons as it does now, but more efficiently.

Better management of polys and some better edge-loops.

Edge-loop the shoulders and biceps. Edge-loop the knees. At the very least. Maybe even edge-loop the elbow/fore-arm, calves, and quads. And maybe, just maybe, fix the gluts a tad.

More polys don't really do much good if you don't know how to use them and know how to get them to play nice with the rig.

Might as well start planning it now, I don't want to be still using the same mesh in 10 years time because some oldbies don't want to make new product lines.
Then again by then the opposition might have bigger sims and sim crossings that work as well as everything else wrong with SL that we can't fix for fear of breaking 2003 content.

On balance I think it would be best to replace the mesh with a better one as soon as possible. Issues with the current one will not go away and we will just have to deal with the problems of broken content, updating the client and so forth sooner rather than later, preferably, because an avatar update is inevitable.

It doesn't make much of a difference whether the new mesh is improved with a rearrangement of existing polygons, more polygons that require souping up the Grid or whatever, so long as the result looks and works better in-world and that it allows for sharper texturing. There is plenty of discussion both in the JIRA and in these forums regarding improvements to the avatar mesh as well as complaints about the limitations of the current model. Logic dictates that it is time to act on lessons learned and perhaps introduce a new mesh on the Beta Grid if that is possible.

On balance I think it would be best to replace the mesh with a better one as soon as possible. An avatar update is inevitable since there are too many issues with the current mesh that are not possible to fix and we will just have to deal with the problems of broken content, updating the client and so forth, sooner rather than later. There are plenty of discussions both in the JIRA and in these forums regarding improvements to the avatar mesh as well as complaints about the limitations of the current model that inform this argument.

Logic dictates that it is time to act on lessons learned since the mesh was introduced and perhaps put a new mesh on the Beta Grid for residents to test if that is possible.

It doesn't make much of a difference whether the new mesh is improved with a rearrangement of existing polygons or more polygons that require souping up the Grid or whatever, so long as the result looks and works better in-world and that it allows for sharper texturing and smoother joints in animations and poses.

The avatar mesh and animations are better than I've seen in any other virtual worlds, apart from those on the SL clones, which are identical.

I know they can't compare what is possible with the best of modern graphics software, but SL has to be able to work online and on various types of computers.


I think our avatars could do with a few tweaks though.

]1) Tidy up the areas where things don't quite work.

2) Add more flexibility in the spine and make joint articulation smoother.

3) Make it possible for flexible prims to be non-phantom, so hair rests against skin instead of disappearing into it and we can have properly fitting skirts.

I'm no expert, but if the mesh is improved, it is still defined by a set of co-ordinates and so I'd have thought that it would be possible to make it accept existing textures properly. Presumably, though, that would mean you'd still have to have the same texure on both arms.

I think it's more important that we get proper dynamic shadows in SL, that would be the single biggest graphic improvement.

The mesh isn't downloaded to the client, it's built-in. So long as the UV map for the "new mesh" is applied to textures with the same resolution the required bandwidth wouldn't change.

The skeleton and the mesh don't both need to be changed at once, so animations should remain safe.