Are we overdue for a new AV mesh?

Reflections would give them a run for the money, especially considering that they had it working so long ago... so they probably wouldn't need a new video card:

I wonder, though, if one couldn't use clothes and skins from the old avatar mesh with a new one, if one only added polys, keeping the existing vertices in place, and running a one-time conversion that projected the old textures onto the new mesh.

To be honest, it seems to me that even if one rearranged the existing mesh, there could be a defined distortion mapping from it to a new mesh. Not the kind of computation one would want to do every time textures were applied, but still, I'm not seeing how this is impossible.

If I'm thinking about this correctly, either way the results would be no better and possibly a little worse than with the current mesh (more worse if current vertices weren't preserved), so there'd still be incentive to upgrade stuff, but at least it wouldn't necessarily obsolete everything instantly.

The other thing is, if we're going to the trouble of defining *a* new avatar mesh, maybe this would be the time to make it possible for users to define *individual* avatar meshes within some poly-count constraint--with or without a defined mapping from some skin-and-clothing-market standard(s).

(And then, if the mesh can be arbitrary... what about the skeleton? E.g., should we make "real" tinies possible while we're at it?)

do eet.

OK dumb question, maybe... let's say that LL come up with a new mesh with a lot better detail on the body. Couldn't they also come up with a "convert" doohickey that lets you go through and convert all your inventory from the old mesh to the new one?

Would something like that be possible?

It would be better to have any conversion done when you bake textures on the client... since the calculations involved would be a drag on the server if done en-masse, but would be a small part of the work going on on the client anyway.

Have Linden Labs ever commented on the viability of implementing a new avatar mesh?

The only thing I can recall is that they said that custom meshes would require too large an asset to be practical. But that was talking about downloading the mesh from the server.

I think it would be useful if they could tell us what practical improvements they could make and when they plan to implement them. A demonstration on one of the alternative viewers would be good for discussion and planning: content creators could decide whether or not it is worthwhile to continue putting their energies into avatar texturing or work on other projects until the matter is settled.

My husband looked at the mesh and said it sucked. It was highly apparent that LL did NOT use a 3-D artist to create the mesh. And in Windlight, the "twitches" show and horribly. A new mesh might actually render face-lights unnecessary.

As a clothing designer, I'd gladly redo my current line up of dresses for a new mesh, if it meant over all "prettier" avatars. I'd gladly rebuy my skins if it meant a prettier avatar. LL might improve user retention if the base avatar was better meshed. In a virtual world, appearance is everything.

Damn right.

As I recall, the avatar we use is based on the version 1.0 Poser Female avatar, which was pretty crude by today's standards. I haven't looked within Poser at the UV mapping on the legacy Poser "Ruth" figure, which they still offer for use as a low-rez background character, but I suspect it's the same bad mapping that we suffer with today in SL.

If they COULD offer concurrent use of a better mapped avatar, even if it had no better poly count, I'd happily migrate, over a period of time, to the new one - even though that will cost me hundreds of real dollars for new skins and clothes. As long as I had the option to gradually obtain and migrate to the new stuff, I'd be fine with it. I couldn't afford to drop one avatar mesh and start fresh with all new skins and clothes, all at once. there would have to be a transition period.

It should be possible to have a utility that will take the current skins and clothes and attempt to remap them for a new figure. DAZ Studio and Poser have such utilities for their much more complex figures, and they are aimed primarily at content creators who would use them to rework their old inventory of products to the new figures. If nothing else, it provides a better starting point than re-creating content from scratch. But it is something you would want to do off-line, and have the new items be new assets. Which means the client would have to treat shapes and skins and avatar mesh as a selectable inventory type.

realxtend is claiming to support avatar 2.0, they allow custom meshes or something like it. I havent played with it yet. I am waiting for the realxtend beta on openlife.

If it is any good, id assume we will see something like it in SL shortly. LindenLabs cant afford to fall to far behind.. on its own technology.