RE: The Animesh Project
I've been thinking about animesh breedables for a while.
The way it currently works, we have one model for a cat which is just a few prims. Each prim is given a sculptie shape and textured. This generally means there's seams where the bits come togetner; ear to head, head to neck, leg to body. For KittyCats, that means a few set shapes. A head, a body, a few legs, each tail trait, left and right models for each ear trait. In all, it'a a managable number. And adding a new ear or tail trait is fairly efficient. Just create one or two models.
Mesh, however, specifically sets out to prevent this sort of thing. You can't change just the tail, or just the ears. So you have to change the entire model. Each new ear trair means a new model for every existing tail trait. Each new tail means a new model for every ear. To make matters worse, we cannot change the model, at all, by script. So, instead of a cat having just a set of prims to which we apply shapes (sculpties) we need a full model for every ear/tail combinantion.
So, mesh makes adding new traits expensive (we have to create so many more models), it also makes it hard. We really don't want to force everyone to have to update every cat just because a new ear or tail has arrived. So some sort of central repository is needed. That means the parents don't actually HAVE to models in inventory. Instead, one of them requests a model from a central server, which delivers the required model for the offspring for the parent to place in-world. Whether the central server sets the traits (colors, etc) before delivery, or the parent gives them afterward is a design choice (I'd lean to having the parent do it, at least initially). The problem is that we have to get the offspring model from the central server to the parent. That presents a single point of failure, which is bad. And it presents a scheduling problem, should a large number of requests appear at the same time (meaning: it can be laggy, too).
My point is that while a non-breedable mesh animal is definitely impressive, and it sure would be nice to do it for breedables, there are a number of hard problems which must be addressed to not only provide a breedable product, but to do so in a fast, reliable, low-lag manner.
|