Create animation videos from 3D joint positions

Question

Sorry if some parts of the question or my followup requests may seem silly, but I am not experienced at all in this field and unfortunately I am on a deadline.

The idea is simple to explain, I have (frame-by-frame) exact 3D coordinates of 25 joints of a Human body. What I require is the corresponding animation video of a human character around it. The idea is similar to this - https://www.youtube.com/watch?v=RPrdDlQXjEg

The difference is that I have frame by frame 3d joint values, not mocap data. Also I want to repeat the process for a huge number of videos so I guess I cannot be doing it manually for every video (again my presumption, if I am wrong let me know).

I have no experience whatsoever in this field. It may seem like I did no research prior to asking this question, but I truely did. It's just that I am really confused as to what my options are and whether it is doable or not.

I don't expect a detailed answer (although if I got one that would be awesome). I am just looking for someone to point me to the right tutorials. Thank You.

Edit : Discussion continued - https://blender.stackexchange.com/questions/122369/create-animation-videos-from-3d-joint-positions

Answer 1

Mocap data is just a collection of points moving in time, data from a mocap system may be formatted better than the data you have, so it could be easier to say these points are the left elbow and these are the right ankle. If you can't specify that this sequence of points is this joint in the armature, you can expect troubles.

Let's start by saying you have a sequence of locations clearly identified for each tracking marker by name. Start by creating an empty that recreates each marker.

import bpy

scn = bpy.context.scene
objs = bpy.data.objects
rig = objs['myrig']

mocap_data = {
                    # each item in list is (x,y,z) at each frame
        'elbow.l': [(0,-.5,0),(0,-1,0),(0,-.5,0)],
        'elbow.r': [(0,.5,0), (0,1,0), (0,.5,0)],
        }

for m in mocap_data.keys():
    o = objs.new(m, None)
    scn.objects.link(o)

Each bone in your rig should be named to match the naming of the markers. In the same loop you can also create a ChildOf constraint for each bone, this makes the bone follow its marker.

c = rig.pose.bones[m].constraints.new('CHILD_OF')
c.target = o

Now you want to animate each marker to follow its motion.

for f in range(len(mocap_data['elbow.l'])):
    for m in mocap_data.keys():
        objs[m].location = mocap_data[m][f]
        objs[m].keyframe_insert('location', frame=f+1)

If you want to, you can bake the animation into the armature so you can remove the empties.

Blender includes the Motion Capture Tools addon, which provides tools that may help at some stage. You may also want to look at the Carnegie Mellon University Mocap Library Browser addon, this addon downloads mocap data and assigns it to rigs, it includes a way to transfer the mocap data to a MakeHuman rig, so there may be something you can use or adapt.