adminspotter
commented
8 years ago Instead of "frames", we should do "sequences" or something like that. To animate a character (or anything else, really), and not make it look like a soldier from Quake 1, I think we'll need some animation sequencing, rather than the keyframing we originally had in mind.
Perhaps the XML will contain some sort of programming for each animation sequence. We can parameterize movements, so if they're small, we can react in a small way, and if they're big, we can react to the maximum extent. Something like the control surfaces on an airplane would be doable with such a parameterized function. Walking and running for humanoids may not end up as the same movement sequence, but these could also fit the concept - a very slow walk and a very fast walk use a similar set of movements.
Our client program currently receives points, which it turns into boxes with geometry shaders. We can continue to receive points, but we want to make them into actual characters or ships or rocks or whatever. Not sure if we'll want to continue with geometry shaders, or do VBOs with the actual vertex data, or what.