Monday was a busy day. Lots of mesh related things going on. The Content Creation/Mesh Import User Group is where most of the news breaks. This week we were supposed to find out where the Lab was in their research for how to handle mesh clothes and body shape issues. There is lots of information on the subject.

New Tree

Charlar Linden has changed from his deciduous look to an evergreen look, which is probably seasonal thing for well dressed trees…

Unconnected Vertices

An unconnected vertex is a vertex without an edge connection to any other polygons. They are somewhat like crumbs. They usually serve no purpose as there is no way to render them. So, the question was asked what the SL Collada Uploader does with them.

Prep Linden says they are not rendered. If one has unconnected vertices the uploader should alert the user to the problem. In general you don’t want extra vertices driving up your Land Impact costs. Prep asked we file a JIRA feature request. I am not sure who is filing it.

Pivot Points

If you have ever made a door in SL, you have probably worked with pivot points. In Blender you can set an objects pivot point. The problem is the SL Collada Importer does not upload the pivot information. The work around is to use an unconnected polygon to adjust the bounding box and thus the pivot point. SL puts the pivot point at the middle of the bounding box. The cheapest adjustment is a 3 vertex poly. Typically two of the vertexes are at the exact same coordinate, thus making and invisible polygon.

So, the question comes up as to whether filtering out unconnected vertices will wipe out those unconnected polys.

Prep expects the use of an unconnected polygon to result in broken objects when the Lab does something with pivot points. So, the trick of using an invisible polygon is a bit shaky and may eventually result in broken objects. Prep knows more than I do about how this stuff works. But, if the viewer at some point stops uploading unconnected vertices and polygons I’m not sure how that would affect previously uploaded objects.

Vir Linden thinks ‘degenerate’ (unconnected) polygons are already filtered. But Motor Loon says he is using them now and they are working to control the pivot point.

Prep says there currently is no Viewer support for changing the pivot point. But a few months back they did start coding for it. Some older viewers did upload the pivot information, or so some of us think. But, there is no way to change it once uploaded. The feature addition to the viewer is currently in the backlog. So, we may see it someday.

In the mean time there is a work-around for pivot points. There is even a video work-around tutorial.

Skeleton

The question was asked, ‘Do meshes use the same exact skeleton system as a normal avatar, even though the joint names are different?’ The skeleton is the same, per Prep Linden. However, joints can be offset.

I’m not clear on this. But, it seems the default avatar uses one skeleton and the skeleton used with a Collada avatar import is a different but identical skeleton. Prep is saying they are identical, but some names are different… or accessible… or something.

Is the joint orientation and rotational order adjustable, or is it pre-determined in the source code somewhere? Prep Linden answered, ‘We don’t support the importing of joint rotations or scaling, atm. (it’s something that I have finished on the importer and just need to get working on the inside). Which is something that I have in my backlog.’

Echos Remblai asked, What is the rotational order in BVH files used for by SL? If it is wrong, will it cause gimbal lock, etc.? I’m referring to the bvh file:CHANNELS 3 Yrotation Zrotation Xrotation. The order of them can be changed on the joint.  – This question stumped the Lindens. They will get information to Echos. So, if you gotta know chase him down.

Can you animate the additional joints of the skeleton when using a mesh: mAnkleLeft, mToeLeft, mEyeLeft, etc? Or is this planned? Prep Linden: All 26 joints should work – all my samples have anyways.

The BVH file spec page for SL says it supports constraints, is that actually implemented and a usable feature? Prep Linden: I’m uncertain – I’ll ping Richard and get his take on this.

Runtai Linden posted a mapping of bones to internal bone names for the mesh model importer, it says lFoot = mFootLeft, is that a mistake? The closest bone location to it is actually mAnkleLeft. In the original bvh animations for SL that bone corrosponds to the mAnkle though in the avatar skeleton xml its the exact position for the ankle too. Prep Linden: nope that’s correct (just looked at the code).

Mesh Deformer Update

Charlar says the engineers have their investigation into mesh deformation solutions. …yeah, that told me a lot too. I suppose that means they have figured out what solutions are available that may work with Second Life.

Now they are ‘vetting the options for technical viability and scope.’ Sounds like management speak. I suppose this is where they get the free 30 day download and test the various packages and processes.  (roll your eyes here) They will probably be testing performance and ease of use. Somewhere in there they will likely be estimating what it is going to take to implement each solution.

The result is that while they have moved forward, they are not a place where they can commit to a direction. Nor can they commit to a course of action. So, we have some more words and know they have moved forward, but little more.

One of the things learned is the scope of the change needed. Consider. The base concept is to have clothes conform to the avatar shape. But, does that include breast jiggle? And what about the tummy and butt?

Next, what happens when someone wears a full body avatar and puts on a jacket, I mean a Collada import jacket of course? And if the full body avatar is a gorilla or a bear? Do we allow Collada clothes to be worn in layers like we do other clothes layers? Does the system need to adjust to avatar changes while a mesh is worn or just fit the current shape when put on?

Then where is this all going? What future things will the Lab have to plan to handle?

As one continues thinking about it the issues seem to grow. Failing to think things through could walk the project into a dead end and force a delay as things are done over. As more things are added the projects gets bigger and more complex.

Charlar is clear the Lab will work the process on their time line, which means when a decision will be made is unclear.  Charlar pointed out that the explosive interest in Collada clothes is a surprise as they expected much more interest in full body avatars. The interest in clothes and the mesh deformer has changed the thinking within the Lab.

The goal is to decide something soon and at least set a direction. Hopefully the Lindens will have a decision in 4 to 6 weeks.

The Deformer Project

Most people are aware the Maxwell Graf is talking with Karl Stiefvater and raising money to pay Karl to build the mesh deformer project. Today an hour after the User group meeting Treet.tv aired: We need to talk about Mesh … on Designing Worlds. This was apparently recorded earlier and aired today. Maxwell, Graham Dartmouth, and Jusden Jonstone were at the showing to answer questions afterwards.

You can find the show here: Video. As time passes you will have to scroll down to find the show. This LINK should be permanent, but I’m not sure.

The first 9 minutes of the show is about what mesh is and what problems we’ve run into.

From 009:00 to  Maxwell is explaining what a parametric deformer is and how it works. Basically it makes our Collada clothes conform to the shape of our avatar. Building one for SL is the point of the Deformer Project.

Along about 15:00 Maxwell explains how he and Karl got together and started the project.

About 17:00 the history of the issue with Linden Lab is rehashed, along with how the LL viewer uploads Collada models. It’s pretty accurate. The big point in this section is that the project will make this feature available to third party viewer developers as well as Linden Lab. (And Riven is mentioned, a Myst game) So, this is a big deal for OpenSim worlds too. We may even see this code in use in OpenSim grids before it makes it to Second Life.

23:00 – This section of the discussion is out of date as it was taped 10/11. But, the information is still mostly accurate, nothing much changed since then.

27:00 – This section provides the answer to what happens if the funding goal is not met in the allowed time. Basically, the project will be extended until they get 100% funded. Over half of the funds have been collected in 5 days. The target goal is US$5,400.

32:00 – Maxwell describes what the parametric deformer should do for residents and creators. He lays out some of the additional or future possible projects. Think a sport coat for a dragon.

35:00 – The importance of having been able to raise $3,600+ in five days.

40:00 – Starts Q&A

Can one contribute in world?Using Lindens? It will not be possible to use Lindens. There are ways to use MOAP to access the web page at Indiegogo from inside SL. This avoids having to exchange Lindens and keeps it clear where the money is and who is getting it. If you need a donation board to put out in SL contact Maxwell or make one. A tutorial is here: Second Life MOAP Tutorial. I made one your are welcome to have. The URL for donations is: http://www.indiegogo.com/Mesh-Clothing-Parametric-Deformer-Project

43:40 – What is rigging? I won’t repeat that answer as I’ve been through it in several articles.

44:39 – Will mesh layering add additional alpha layers? Probably not. Part of this project is to eliminate the need for alpha layers.

Deformer Summary

The community is raising funds to pay for the project. Fund raising is doing well.

A complication is the timing. We may have funding in place and be ready to start the project before the Lab has made a decision on direction. I initially thought this might hold the project back. But, pointed out in the video is the point about the Emerald Viewer adding the additional attachment points. It ended up being incompatible with the Lab developed attachment points and eventually had to be changed in the Phoenix Viewer. But, the popularity of the feature in Emerald moved the Lab to change priority and develop their version of attachment points and clothing layers. So, this may be a similar case.

The project is likely to have follow on projects if this one works out well.

 

Summary

We learned to day that the Lab is figuring out how they want to handle Mesh deformation for better clothes and avatars. The Deformer Project is working on funding. And Charlar has a new look.