Ha! pinBone is making great steps now. I have it extracting transform data for a given bone at a given frame in a given action, which is what I need to build the nla transform matrix.
Working on pinBone step by little step. Every time, it gets a little closer. The more I do, the better I understand the internal structures I'm working with. pinBone can now extract the proper actions, armatures, bones, and poses that it needs. Now I'm going to start working on calculating the proper matrix to cache with the NLA strip, to be later used in the transformation section.
Slow but steady.
I've been very busy since my last post, almost a month ago. I do a lot of development on BP and the Blender source at work, while we're slow. We've been extremely busy this month, and my coworker is now on temporary active duty with the Air Force Reserve. And that means no free time at work.
I finally got some time to work on pinBone two nights ago and made some significant progress. I got the NLA panel GUI code back from Matt Ebb (thanks!), incorporating the pinBone field. I now have Blender able to extract the actual bone object given the NLA strip and the bone name. Left to due are bone matrix determination (the next step), and finally the application of the matrix during do_all_actions.
So it's moving, but slowly. I've found that I have to have something like a 3 hour block in order to get anything really done. I can't just grab an hour here and there.
I've also been experimenting with an Aqsis/RIB-aware version of Tuhopuu. The export works well, and I believe that once I learn to tweak the rendering and lighting parameters properly, that I'll be able to get some nice looking outputs. My eventual goal would be to have BlenderPeople's final full-on character animated demo video be rendered with a high-quality Renderman compliant renderer, with tranformational motion blur and DOF.