#SL Mesh Deformer Alternative

UPDATE: 4/2013 – Several designers are using alternate methods to make sizable clothing. The Lab is advising us not to use non-supported features, which these alternatives are. While the Lab will not deliberately break them future changes may inadvertently break them.

Today was a surprise. I tried to go to the Mesh/Content Creator User Group and the region was full. WTH!?! I had to TP into Pooley, next door to Borrowdale the usual meeting location,  and see what was up. In a minute  or so I realized the meeting was being moved to the border between Grasmere and Borrowdale. I flew around and was soon in the middle of a chaotic meeting as Nyx Linden built prims so we could see the region boundaries and divide up.

Surprise Crowd at User Group Meeting (60+)

With 60+ people trying to get into the meeting things were chaotic. I crashed a few times, did a Doh! by trying to take a picture (I felt so dumb), fell through the temporary decking into the sea a couple of times as Nyx moved the deck. Whatever, I survived and my SL hair doesn’t frizz.

The Alternative

By now you hopefully understand the problem the Mesh Deformer is to fix. Our mesh clothes only conform to some of our shape settings. The Deformer makes the clothes conform to all our shape settings. You can only see the Deformer work when you use the Project viewer.

But, there are some mesh clothes out there that conform to our shape and DO NOT use the Deformer. They conform now and the change is visible in any viewer. I’ve been aware of them for 3 or more months. But, the developer was working out details and dealing with some of the downside, and as much as I don’t like NDA’s I was convinced to keep quite. But, now the cat is out of the bag.

So, what is the alternative and how does it work? I’m not sure how it works, yet. I’m figuring it out. But, the process skips using the current bones and does the weight painting on an alternate set of bones one can find included in the viewer files. Such mesh will upload as it is OK to have zero weights on the regular bones as long as the standard bones are included in the upload.

So, by weight painting with the painting assigned to vertex groups linked to the other bone set, things seem to conform to the shape settings. If I understood correctly, there are problems with some parts of the avatar, like breasts. So, the Lab will need to add bone definitions to make this method complete. But, adding bones to the alternate bone set in the viewer avatar definition will not break existing content. So, it seems an easy fix.

The Story

Since I don’t really understand yet I’ll quote Redpoly Inventor, the guy that came up with the medtod.

First the trigger for the large turnout was this agenda item:

Can Linden provide any feedback/knowledge concerning a method of mesh clothing upload which circumvents the use of the existing mesh clothing deformer yet provides similar (if not the same) functionality? For background, a known full-perm clothing merchant recently made available an article of clothing which conforms to changes in av slider values. ( http://www.youtube.com/watch?v=POwQE10sxGg&feature=player_embedded )They claim the item does not use Qarl’s mesh clothing deformer but decline to discuss the method being used behind the object creation. Speculation is that a hack is being done to .dae files in order to get this to work. If this is the case, is this a method of content creation which designers should or should not make use of at this time? Aki Shichiroji

I suppose that is what got the Fashion Design group to show up. This is all new to Nyx and Oz Linden apparently.  Now in Redpoly’s words with a considerable amount of my editing:

I’m here for that 🙂 to explain in detail what and how.

the trick is to use the collision volumes for the rigging part. they were designed to scale and move with the morphs in order to cover them. I assume [they are] used for physics [at] some [point] time. the problem with them for the rigger [person rigging] is that they have special rotation, size , and position. [You can find them] all listed in the avatar_skeleton.xml. [You can select] show collision volume in the viewer [and] wear [an] alpha [to see the …whatever.] [When you change your shape] you will see them scale up , and move. they move and scale in a very specific way, [which is] all listed in avatar_lad.xml.

each appearance slider triggers not only morphs , but volume morphs too. So, [it is] all good, but there are tons of problems and down [sides], let me explain.

1st you need new skeleton linked to the mBones , created based on the avatar_skeleton.

2nd you need to have this movement (avatar morph + volume movement/scale ) simulated in your software of choice. Without this [movement] you will work completely blind.

If you have been doing weight painting, you know you need to move the avatar and change poses so you can see how the painting is affecting the movement of the object. Imagine trying to do you painting and then importing to SL to see how it did or didn’t work. That is a big downside. Red is hoping Gaia Clary, the Avastar person, can solve that problem.

I have another proposal based on this, which will work much better. i’m talking about new set of bones. like to call them cBones. they will be linked to the mBones [at] 0 rotation , 1.00 scale. but sliders in the LAD file will trigger movement , improved one, and scaling. because with collision [stuff], the belly part is very very problematic. many clothing models can’t be made to work with it. u need to hide it big time with alphas.

my initial tests on the cBones idea indicate that we will need second torso. to cover 100% fat at the back and the belly. we will need 2 bones for the breasts and probably 1 or 2 for the butt. some of the bones , mBones are already scaleable, like mHead , mHands etc. so they don’t need to be cBones.

The control over the deformation will be big and those will be [able to be animated] which can be very beneficial.

[The] collision volumes are [a] solution, but a very limited one [that] triples weighting time. Without a programmed simulation system [in the modeling tools] it’s almost impossible to work with.

Nyx Linden responded, “Sounds interesting, and very challenging.”

RedPoly Inventor: Nyx the cool part is patching this is… plain text 🙂 [so] all can work on perfecting avatar_lad [and] skeleton.xml [files].

That’s all.

Nyx summarized, “So, sounds like there are two different discussions here: 1) the viability of using the existing collision volumes for weighting, and the drawbacks there, and 2) the proposal to add more ‘cbones” that would track the morph params more accurately than the current collision volumes.”

Summary

This is going to take some time analyze and experiment with. Over the next week we can expect more information. People like Gaia Clary, Elie Spot, and Ashasekayi Ra are looking at what they can do with the new ideas.

Expect more interesting news soon.

Updates ——————————————————-

A demo can be found in world in Daxos.

A discussion can be found at SLUniverse.

15 thoughts on “#SL Mesh Deformer Alternative

  1. Pingback: Second Life Mesh Clothes Blender 2.6 Setup 2012 Tutorial

  2. I was afraid of something like this.
    The longer LL waffles around with implementation of a method for making mesh clothes fit, the more clever people will find hacks and alternative methods of accomplishing it — and the longer it will take for LL to make up their minds and do *something*.
    Of course if mesh clothing had even occurred to the geeks in the back room we would not be seeing this mess at all.
    I did a search on marketplace for just “mesh”. It turned up 90% or more clothes even before they really fit right.

    • Judging from the clothes they brought today, this process works better then Deformer 0.3. Karl’s next version using normals in the calc’s will probably do as well.

      Also, this idea was mentioned as proposed by Prep Linden, or something very similar. So, it is not exactly something new nor is a hack. I’m hoping we can add a couple of bones and links to the avatar and have clothes work much better.

  3. Pingback: Alternative to Mesh Deformer? - SLUniverse Forums

  4. Sounds simple but at the same time a lot to take in. Thanks for the info. Can’t wait to see what Lindens and others have to say once this is looked at more.

  5. Pingback: Mesh clothing deformation: alternative approach suggested | Living in the Modem World

  6. That’s the best I was able to do, Nalates. I was shocked, like you, when I go there for office hours.
    http://drfranbabcock.wordpress.com/2012/06/25/crazy-mad-scene-at-nyx-linden-office-hours/

  7. Pingback: Mesh Deformer 0.3x Limits

  8. I’ve only dabbled in Poser, but I’ve often had the feeling that the Lindens are hopelessly out of date on the subject of Mesh. Some commercial Poser figures have, since SL started, gone through 5 generations. They’re just too complicated for a real-time environment such as SL, but the standard AV Mesh feels horribly old to me. When the Deformer project revealed some of the problems with the AV Mesh, I began to wonder why it had never been updated. It would be possible to keep the UV mapping of the original. It need not have been a V2,

    • Much of the problem with the avatar is in the UVMap.

      If you think SL avatar looks primitive, check out Cloud Party’s avatar.

  9. Pingback: It Almost Looks Good | MetaReality Podcast

  10. Pingback: MetaReality #SL Mesh Deformer plus Cloud Party

  11. I think that this discussion is healthy and timely. It’s something that certainly could have been brought up before, but as RedPoly says, it’s not really a great alternative since the particulars are very tricky and simulating in the development tools like Blender/Maya/etc. can be difficult.
    I don’t think talking about it is waffling. It’s more analysis, and it’s a lot less expensive to do that before the deformer gets released.

Leave a Reply

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