Ebbe does say they will be trying to automatically optimize content being uploaded into Sansar. If you have tried to let the SL viewer optimize mesh content, you know what a mess that is. Everyone I know recommends creating your own optimized models rather than allowing the viewer to do it.
If you have used the poly reduction tools in 3D Max, Maya, and Blender to create low poly models automatically, you know how difficult it is to get good optimized models. In general it is still a manual process. So, the Lab is taking on a task no one has successfully accomplished, IMO.
That Ebbe is seeing things never seen in SL is not surprising. We expect the animation armatures to be much more flexible in Sansar. So, 8 legged spiders would be much easier to make in Sansar. But, are they better optimized? I suspect so. After all the developers now in Sansar were handpicked by the Lab. I doubt they picked people known to develop high poly hard to render products.
All of this still leaves us to wonder what style of rendering the Lab is building into Sansar. We only have hints. We can see High Fidelity is going for a more cartoonish avatar. Philip Rosedale, HyFy CEO, sees the Valley of Uncanny** in a different place than I do.
In Second Life we have more and more realistic avatars. SL users may be becoming adapted to tolerate a limit farther into the ‘uncanny’ limits than most. But, how many SL users are going to want a cartoon avatar? How many SL users will want a Halo styled world? Or any world less photorealistic?
For general marketing use by RL businesses outside the SL-type worlds (Ford, DKNY, …), how many will want to advertise their products in a ‘stylized’ world?
I suspect we will be able to create highly photorealistic experiences in Sansar. I expect the Lab to cater to the existing SL audience. But, we are not their only target demographic. But, I believe we are a well known and important part of their target audience.
*Baking in Lighting – this is technique that is used in games to reduce the render load. In SL we see this used sometimes. In Blender we may bake in the lighting effect for a model. This composites the shadows and highlights into the texture for the model. The shadows and highlights do NOT move with the sun or lights. To look more realistic we can omit the shadows and highlights and let the SL sun cast shadows and highlights. In VR the shadows will need to move with the lighting just as in RL to maintain an immersive environment.
**Valley of Uncanny – There is a point where human looking avatars get spooky. They look real enough but various hints tell us they aren’t and that spooks people and they avoid them. A less real looking avatar can emit those hints and not be spooky and people accept them. So, there is a known range of limits (fluid and different per person) to how real an avatar can look. The range in which people find them unacceptable is narrow. So, avatars have stay below a limit or exceed it by a significant margin. The range of unacceptable is the Valley of Uncanny. Think Burger King’s freaky King and Jack-in-the-Box’s comical Jack. Which do you like?