RobynsVeil opened this issue on May 30, 2008 · 267 posts
moogal posted Wed, 18 June 2008 at 4:32 PM
There's another thread going on currently about the future of "content" figures. It's largely a discussion of advantanced and klugey rigging techniques and how long it will be feasible to support both the Daz and Poser markets with the same characters.
There is a clip in the video I linked to above where a figure is posed simply by grabbing and dragging the hands and feet. The neat thing is that this is not a rigged figure, but a solid mesh. They are using a custom solver to somehow discern (via body mass?) how this character should bend. It's not a totally realistic effect, as the character ends up moving like a rubber figurine, but at least to the eye it looks good, especially with the mo-cap data applied.
This got me thinking that the future really isn't in the figures, but in the rigging itself. Mesh resolution and rigging are the two areas where the most attention has been paid to commercial figures, and with each new one comes the need to build or buy complementing accessories.
So, my idea is similar to zBrush modeling rigs. I'm thinking of a kind of default rigged bipedal cage. It has the proportions of a basic human, and has been rigged down to the hands and feet, perhaps even to the fingers and toes. What it would do is let you import a figure of your own creation, say from makehuman or zBrush, and then adjust the cage until it fits the imported mesh. Then the mesh could be bound ot the cage and a new figure would exist, able to share any pose or motion created for the pose rig. Muscle deformers could be linked to the pose rig and would available for any future figure created. Of course changing the size and proportion of the pose rig would affect its poses. (Perhaps a special-case use of IK could try to make sure that hands and feet stay in the same absolute locations, as they are the parts most likely to be in the wrong place when using one figure's poses with another.) A more advanced solution could perhaps try to make sure parts didn't intersect, say when a pose for a skinny figure is applied to an obese one. Also, shouldn't it be much easier to transfer facial poses from one figure to another than it currently is? Why can't I just line up the new figure with the old one and have the software create corresponding morph deltas? Maybe someday makehuman will be able to let me match the default figure's proportions to an imported mesh, then transfer a pose from the default figure to the imported mesh? It'd be a start, like conforming in reverse almost.
I'm sure most of this is off topic. But as I keep coming across threads prediciting the end of Poser, I can't help but think that it's the figure business causing the problem. There might not be a way to make rigging automatic, but for a basic human figure it shouldn't need to be done from scratch every single time.