Mesh clothing deformation: options

The informal content creation user group is into its 2nd week of meetings. I attended the first, but frequent Viewer crashes meant I lost a huge amount of context during the meeting (particularly as a non-specialist), and didn’t feel entirely comfortable providing a write-up based entirely on the transcript of the meeting. I also missed this week’s meeting due to working on other things.

However, Nalates Urriah provides an overview of the meeting and the core discussion on mesh clothing deformation, which I’m using, together with the meeting transcript, to provide a summary here.

As most people are aware two deformation options have been put forward: the parametric deformer first proposed by Max Graf and which Qarl Fizz has been developing (and versions of which have been made available through an SL Project Viewer), and recently the option presented by RedPoly Inventor.

An early video from Qarl on the parametric deformer

For content creators, both options have certain advantages and disadvantages, and opinion on both has been somewhat split. Of the two, however, the parametric deformer is furthest along in development and potentially offers the more direct means for creators to ensure mesh clothing fits. While RedPoly Inventor’s system offers the advantage of appearing to work with the existing shape sliders, in its current form it will be reliant on alpha layers to a far greater extent, adds complications to the process of weight painting mesh, and would likely require the development of a new set of avatar bones in order to fully work and meet consumer’s expectations.

During this week’s meeting, discussions continued around another “alternative”, that of using morph targets – a special shape that defines how a mesh should deform when a certain parameter is increased or decreased. Morph targets are widely used in 3D modelling and are in fact already employed in Second Life to some extent: they are activated when using the existing appearance sliders, However, in order for them to work in terms of mesh clothing, etc., would require updates to both the viewer and on the server-side and would be dependent upon additional data handling – so further support from Linden Lab would be required in order for this option to mature.

However, it does offer advantages:

  • It is theoretically far more flexible than either the parametric deformer or RedPoly’s proposal
  • It would leave content creators entirely in control as to how a mesh deforms
  • It would probably offer faster rendering than Qarl’s deformer
  • Morph targets might be used for animating mesh avatar facial features

For those interested in the more technical discussion on morph targets and SL, the transcript of the meeting is the place to visit, covering as it does such diverse aspects as LSL support for the approach, working with avatar physics and so on. In particular, there is detailed discussion on what needs to be put into any proposal relating to the use of morph targets that could be put to LL – a discussion liable to continue in next week’s meeting.

As Nalates points out in her article, there was also discussion on how to move things forward vis-a-vis Qarl’s deformer – Qarl has previously indicated via Metareality that he is more-or-less waiting for a consensus from “the community” (although it is doubtful any consensus can be reached without LL having a cast of the dice). This also involved debate as to whether things have to be an either / or solution, or whether things could move forward on more than one front.

To this end, Nalates has set-up a poll on her blog. If you understand all of the issues involved on the situation, and have a clear opinion on what you would like to see, please take a minute to complete the poll.