SeanSteel wrote:No new vids yet, I've gotten alot of work done improving the skeleton. The pelvis has been at the wrong angle forever. I rotated that backwards almost 45 degrees! I made the shoulder blades more realistic, and positioned them better. They where a little far apart. Also the arms and legs are new. In the beginning I was trying to entire arms and legs in AR, the more complex a component is, the more difficult it is to get it right. Like having an entire arm that bends, is more difficult then just adding a couple muscle forms to an real existing arm. So now I'm trying to "augmented" the existing arms and legs. We'll see. I'm trying to get multiple marker costellations working on the software side. Does anyone here do any game programming. I'm using OpenCV in Unity to track markers at run time. I have some ideas if anyone wants to chat.
This is one of the best pics yet of the real model. It used the spine in the above pictures. The "blank" areas in the arms and legs are supposed to get AR visuals too. Hopefully someday.
This is interesting.
Out of curiosity, have you though of using IR detection for positioning instead of reading of symbolic projection?
There would be 2 ways of handling IR detection.
One way is using 2-3 external cameras with Infrared capacity (cost approx. $10 each) and putting 1 IR emitter on each part you want to be oriented in AR. That would be similar to how the Vive VR headset uses its 2 black box to track the positions of the headset and, if applied, trackers in 3D space.
The other way is reversing the camera and IR emitter. Putting up to 6 cameras with IR capacity on the doll and detecting 3 or even 4 IR emitters set in the room's corner. From this, you can position the doll not only in 3D space, but also in orientation. If you cut out the feed of the camera so that it only read colors close to IR, it's actually quite energy-conservative. If you force its feed to be in low resolution so that the IR emitters appears as 1-3 pixels at most, the CPU usage is rather low. 6 cameras would be the best since it would allow you to generate a cubemap (6 sided cubes) and detect the position in those 6 sides of each IR emitters set in the room. Basically, making the doll, the "center" of the world in terms of AR.
Then you can use either the on-doll CPU/device, a 2nd micro PC or directly feed the camera feed to Unity and manage to position the head and hands properly.
Hope this give you some ideas.