Avatar pic by Ella Pinellapin
Continuing his blog series on what he's learned as a lead developer/consultant at Oculus, Eve Online, and of course Linden Lab, Jim "Babbage" Purbrick has a new post on the importance of avatar customization tools in open-ended virtual worlds:
If you are building a Metaverse which connects every possible virtual world how do you start? How do you please the person who wants to have the perfect tall dark stranger avatar rendered in 3D as well as everyone else? This was the challenge facing Linden Lab when building the avatars for Second Life at the turn of the millennium. While the default avatars could be manipulated into a huge variety of shapes and sizes using the available parameters, the results were often crude and the available wardrobe never strayed far from geeky teen hangout chic. Luckily for Linden Lab the inability of the avatar system to satisfy every whim was a huge opportunity for the entrepreneurs of Second Life.
Builders quickly realized that they could fashion shoes, hair, clothes and accessories from primitive 3D objects which could be attached to avatars to dramatically increase the gamut of styles that could be realised. Linden Lab enabled another wave of innovation by allowing animations to be uploaded which as well as allowing every conceivable dance move to fill the nightclubs of Second Life also enabled tiny avatars which either folded up the default avatar geometry inside cutely sculpted animal models or moved the default avatar under the floor so it was completely hidden and just served as a platform for a tiny replacement avatar standing on its head. Many years later Linden Lab finally allowed people to entirely replace the default geometry and textures so now it is common for not a single triangle or texel in a Second Life avatar to have been defined by Linden Lab.
But If a new virtual world is just launching, what avatar options should be available for early adopters? Jim's advice:
"In my opinion developers should make sure they provide realistic body shapes (not 2 meter tall avatars with impossibly large breasts) and diverse skin tones and facial features so that every potential user can easily build an avatar that looks like them if they choose. Some basic clothes should be provided, but these can be largely left to the community to develop."
Adding non-human avatar options open up some additional, interesting concerns. As Jim puts it:
"Non-humanoid avatars are technology dependent: it's easy to describe yourself as a dragon in a [test-based] MOO, harder to animate a dragon in SL and almost impossible to drive a 6 limbed dragon in VR when you have 2 tracked hands and a tracked head.
"I've had several amazing conversations with [influential game designer/Journey producer] Robin Hunicke about this: as technology improves your ability to be someone or something different in a virtual world is lessened. Once you get to tracked VR it's like you're wearing a dragon Halloween costume. The infinite potential of the virtual world is constrained by the limitations of the real world."
In other words, as graphics and animation quality improves, there's less left to the imagination.
"VR tracking technology also has a big impact. It was incredibly exciting when people using the Oculus Toybox could recognise each other by movement alone, but also limited the traditional power of virtual worlds to allow you to take on a new identity and be someone else. Even if your avatar looked completely different your movements could give you away."
My own advice: While you should definitely encourage users to customize their avatars, you also want to impose optimization restrictions so avatars don't become a bloated lag-inducing disaster like they have in SL.
Leave a comment