#SL Content/Mesh News Week 19

There was no scripting user group meeting this morning, Monday. Kelly was away. So, no news on what happened in scripting or server updates.

Strange People at Mesh-Content Meetings

Several questions came up in the Mesh/Content Creation User Group. Plus some problems and tips for saving Prim Count.

Mesh Textures

You may know that over time the number of textures used in a mesh and how they degrade by LoD level active, has been an on and off problem.

Mesh-prims are allowed to have 8 textures or faces. Prims can have one texture per face. With prims the faces are pretty obvious. The faces are far less obvious in mesh items. The creator has control over what is a face in their mesh object. The limit is 8 faces. Each can have its own texture.

When one decides to use 8 textures on a mesh items top level LoD (Level of Detail), they must also use 8 textures on each of the lower level LOD’s. At various times that has even extended to the physics level, but that is classed as a bug.

Ashasekayi Ra noticed that in the ADITI grid that the viewer’s upload warning of “different number of texturable faces” could be ignored and the upload would proceed. One could later upload the texture and apply it.

Nyx Linden responded saying, “I think the intention was that lower LODs could use a subset of the materials from the highest LOD, but I’d have to double check the code to be certain. I’m not sure if it’s working correctly at the moment, so reproduction info is always appreciated. I’m pretty sure the current implementation is not matching the intention at the moment.

If you are seeing this error and have a model that reproduces the error consistently, send it to Nyx and let Nyx know what the example is about. Or add it to the Jira.

VWR-28612Incorrect texture-face associations when low LOD meshes have subsets of high LOD materials.

Mesh Attachment LoD

When we wear mesh clothes the mesh uses the same LoD as the avatar. However, mesh attachments that are not rigged mesh items use a different LoD and change LoD at different distances from the avatar.

Nyx Linden  said, “If  I had to guess, I’d say that the attachment is likely switching LODs based on its size, like a normal in-world object, but rigged meshes are switching based on the size of the total avatar.

I never did hear whether that could be changed or whether the Lab would consider changing it.

Avatar 2.0

I’ve been hearing more discussion about the sort comings of the current avatar mesh, the basic Ruth/Roth mesh that makes up our avatars. I asked to find out if the Lab was planning any changes to the avatar. Avatar 2.0 is a nebulous idea that means different things to different people. It is the label we hear applied to almost any change to the base SL avatar. As far as Nyx knows, there is no Lab project looking at changing the avatar.

There is a considerable concern on the Lab’s part to avoid breaking existing content. In Linden eyes content in this case is: clothing textures, animations, and attachments.

Adding vertices to the avatar would fix some problems with the avatar and how it deforms. That will also fix some problems with the coming Mesh Deformer. However, such changes necessitate a change the avatar’s UVMap. Everyone in clothing knows the UVMaps as the clothing templates, like Chip Midnight’s templates.

When one starts to work with mesh clothes they get into vertex weighting. We soon learn there are deficiencies with the existing avatar’s vertex weighting. That triggered a JIRA and discussion about changing the avatar.

STROM-1800The vertex weights of the default character mesh could be better. Drop by and click Watch to show your support. This is not the end all be all for avatar change. More about JIRA’s later.

Nyx says, “…some creators could rely on very specific behavior of that poor weighting. It may be possible to make good tweaks, but it’s just something to be cautious of.” I think this shows the concern the Lindens have to avoid braking content.

Both the vertex and UVMap changes could be made carefully and avoid serious changes to how legacy textures would render. BUT… many consider such changes way too much work for the benefit gained. I tend to agree. But, small changes are possible and can conceivable avoid breaking everything.

Those working with rigged mesh realize a number of problems and deficiencies in the avatar skeleton or armature. Attachments and animations depend on the armature being what it is. Changes to it will affect many attachments and animations.

The existing content in the form of clothes, attachments, and animations is a huge investment of labor and money. People have tens of thousands of items in inventory that would be affected by a change to the avatar. So, any plan to change the avatar is going to require some transition path from existing content to new content.

The Alternative

The Lindens seem more open to the idea of adding things than changing things. I am sure the risk aversion to breaking things is seriously built into the corporate culture of the Lab.

We now have mesh avatar replacements. They lack hand and face movement. Many feel if they can get the Lindens to add those features, people will start to use more replacement avatars. This would allow creators to build very advanced avatars that have much better movement and flexing. It would also allow much better UVMaps that would display clothing and skin textures for better than the current avatar.

As face and hand movements are added other deformations could be added too. Also, access to more of the avatar controls like the jaw, tongue, eyes, and eyelids. With these controls users would have the tools they need to build their own avatars and animate them.

Some creators want to build spiders, 8 limbs. Others want to build snakes, no limbs but lots of body joints. This will require custom skeletons/armatures. This makes it complicated to handle attachments. It can be done. But, an all purpose skeleton and collection of bone names that can be used with attachments has to be devised. All the new stuff has to be in addition to the existing skeleton. Or perhaps a skeleton can be another appearance item like a skin or shape.

Figuring out how the clothing Deformer will work with a snake wearing a T-shirt will ne novel. But, that is what we are asking the system to be able to handle. Otherwise, the user has to know not to place a T-shirt on a snake and that increases the learning curve, not something the Lindens will like.

As a community we won’t move past what we have until we figure out some way to build a user consensus that can address Linden concerns and convince them to spend the time and money… or accept something we build.

Another viewpoint is expressed by Shug Maitland in Re-assessments at the Highest Level. I agree with many of her points. At some point dealing with legacy restraints is simply too restrictive. Technology advances. I believe the Lindens are acutely aware of the problems both Shug and I point out. To imagine a business the size of Linden Research does not have a forward looking plan reaching out 5 to 10 years is difficult.

Leave a Reply

Your email address will not be published. Required fields are marked *