Open jeffcrouse opened 12 years ago
Thank you! Jeff!
I'm sorry I haven't consider about that inconvenience. :P The main reason of the change is that you guess. There is another (small) reason that reducing user's code. In latest version, users don't need to judge each skeleton is tracked or not.
Maybe the solution will be like these below... a. Add an argument (int[]) to getSkeletonPoints() that would be an array of valid playerIDs. b. Add an function that tells which skeletons are valid.
Please let me know which is better. Or any other ideas?
I am very happy to know that our library is used in such a great and beautiful project! Thank you again for using and telling!
Hi Jeff, : ) yeah, recently nariaki joined this project and got some cool updates. some updates may have inconvenience because of the lack of our communication. Thx so much for informing. we will fix them asap. : )
BTW, i checked your project in advance, it's awesome and beautiful! Congrats. : )
In previous commits, you could use the indices of the skeletonPoints array to track skeletons across frames since each skeleton would stay at the same index. This was a very handy feature. For one thing, it let you associate it with a player silhouette ID. It also made it much easier to determine when a skeleton leaves the interaction area because you could say, for instance, if skeletonPoints[3] hasn't been valid for >2 seconds, it's gone.
The current getSkeletonPoints method disassociates the skeleton from its place in the kinect::nui::SkeletonFrame::SKELETON_COUNT and puts it at an index based on how many valid skeletons there are. This seems like a significant loss to me.
I think that this was possibly done so that invalid skeletons aren't recorded. I have a few ideas for solutions, but are there any other reasons? Is there any reason why the skeleton's ID shouldn't be used for this kind of tracking?
BTW: we used your addon for this project: http://www.creativeapplications.net/openframeworks/paik-times-five-by-flightphase/
So thanks!