SLCC: The Future of Mesh


  • Charlar Linden
  • Runitai Linden

Historical context given by Runitai

“So before mesh, the only way to build anything in Second Life was to use prims. If you’re a builder, you’re familiar with the prim system and how difficult it was to cobble together anything that didn’t look like it was made out of Legos.

“To combat that, with as small a platform change as possible, we introduced sculpted prims a few years ago, which sort-of let you use a very loss-y compression mode of making a deformed sphere-mapped mesh. A sculpt is basically a mesh with very little artistic control over what actually goes into it.

“Now, with mesh import, you can actually use any industry-standard 3D modelling tool to create 3D content and import it into Second Life….What this means is anybody who is a classically-trained 3D artist can come in and start using the tools of the trade to make content in-world. You’re not limited by any toolset that we provide; Linden Lab doesn’t have to devote resources trying to compete with companies like Autodesk which makes 3D Studio and Maya, and the creative space, therefore, is not bound by the Second Life platform itself, it’s bound by anything that you can imagine and that create with any external programme available to you.”

Main Presentation – Charlar Linden

  • High-level bullet-points for  mesh defined as the abilities to upload them, see them, interact with them (touch, move rotate, link, texture, etc), there are additional Havok physics attributes and 64m prim sizes
  • Re-stated that PE (Prim equivalency) is a measure of the resource-impact a mesh object has on the land that it is on
  • Prim count, which everyone is familiar with, has no relationship to PE “You can have a tortured torus…that counts one prim, but it really doesn’t. It’s really actually much more expensive [resource-wise] than a little 1 x 1 plywood cube – or, for that matter, a fairly elaborate mesh.”
    • Believes it is up to LL to better inform users as to the real cost of objects on land, and that mesh starts to show this
    • Outlined the major resource weights: network bandwidth (how much information is being transmitted and to how many people), server resources (what the server has to do and run (e.g. scripts)), and physics (the impact on the Havok engine, collision detection, etc.)
  • Commented simply visiting 3rd party websites (used Turbosquid as an example) and uploading objects directly from them, saying that while it can be done, the results will be inefficient and “you’re going to get stuff that look really poor, it’s not going to look very good initially, and as you back away from it, LODs go to crap” because the system cannot determine how things are supposed to look without assistance and, “that’s where you guys come in,” as LODs, etc., need to be worked out in order to import successfully &  efficiently into SL
  • SL currently supports COLLADA  1.4.0, but 1.4.1 being worked on (but may already work)
  • No expectation or desire to see prims rendered useless due to unique capabilities, including the in-world collaborative opportunities for building & are central to the creative aspects of SL
  •  Mesh is viewed as better than sculpties in terms of defining custom UVs, LODs, etc; however,  would like to see focus on “filling in the blanks” to improve mesh where sculpties are currently “better” (presumably in terms of usage)
  • Admitted some work on rigged meshes/ avatar-related features  have been deferred from the initial roll-out in order to ship mesh, but LL acknowledge the requests and that they will be dealt with “later”
    • During final Q&A, stated arbitrary skeletons would be “cool” to have, but is not a “small bit of work” so LL “don’t know yet”
  • Re-iterated that mesh isn’t some kind of “tremor” that is going to come through SL and everything is suddenly going to look better; it is a matter of content reaching the marketplace, etc., and people buying things & create experiences
  • For non-professional content creators, suggested trying Wings3D, and commented LL would like to encourage people to try out 3D content creation whether or not they intend to sell it
  • Technically, LL will continue to work on mesh, and there will be “at least one more non-trivially sized set of improvements”, a combination of feedback from Beta testers, discovered bugs and features LL would like to include, to be “released in the very near future.”
    • Indicated it is likely the uploader UI will be changed to make it easier to use and understand, admitting that some people get poor results with uploads because, “We’ve confused the hell out of them with the uploader”
  • Encouraged people to use the mesh forum for questions, feedback, etc.
  • Demonstrated mesh in-world, showing 48-PE vehicle scripted as driveable & with colour-changing options, etc; a 52-PE house with texturing, floors, rooms, windows, working doors; a 3-PE scripted (with animations)highly-detailed rifle;  a 55-PE tree house as well as work by other creators and artists
  • Indicated that ideas around providing marketing / showcase assistance to demonstrate mesh to non-SL 3D artists / creators have been “floated”, but there is sensitivity about being perceived as helping some SL creators and not others, and that such activities are not something LL do well, but will take into consideration
  • Responding to a question on DAE/COLLADA support, Runitai stated the specification is huge, and LL’s focus in on supporting a subset – how applications like Blender, Maya, Wings 3D, 3D Studio and Zebra., export to COLLADA & ensuring the SL importer works with that subset
    • Indicated that Google Sketchup’s exporter does not create content that is efficient for the SL importer
  • Indicated that the upload fees will be increased, but will not be to the L$150 x PE that was discussed during the beta programme. Fee seen as a means of “throttling” uploads
    • Indicated a DD is available for testing as should be used to avoid being repeatedly charged when testing a model
    • Better previewing mechanism is something LL may come back to, depending on how well the DD mechanism works out.