Ankle Angles

The answer is that there are several difference between the standard ankle computed by Visual3d and the ankle angle computed when Visual3d exports to OpenSim. The difference includes:

1) A standard Visual3d ankle is based on foot coordinate system that has a long axis running from the distal end of the foot to the ankle center which causes about an angle of 70-75 degrees when the tibia is vertical and the foot is flat on the floor. Visual3d must remove this 70-55 degrees of offset so what it does is computes the angle of the foot to floor during calibration and subtracts this value at each frame as it exports to OpenSim. This explains the 70-75 degree offset between your Excel graph 1 and graph; howwver this does not explain the differences in range of motion you notices. The following two notes will explain these differences.

2) A second difference between the standard Visual3d ankle angle (your Excel graph 1) and the angle angle Visual3d exports to OpenSim (your Excel graph 2) is the for a standard Visual3d ankle angles the position and orientation of the foot and shank are computed using the segment optimization (Spoor and Veldpaus, 1980) whereas in the OpenSim export the position and orientation of the foot and shank are computed using Global Optmization (Lu and O’Connor, 1999). Lu and O’Connor (1999) does an excellent job of explaining why the two techniques return different results.

3) Standard Visual3d joint angles are based on a Euler sequence of X-Y-Z rotation where X is directed along the X axis of the shank, Z is along the long axis of the foot and Y is perpendicular to X and Y. The Visual3d OpenSim export is actually a two degree of freedom ankle joint that allows only rotation about a flexion extension axis and a subtalar axis. (Again the amount of rotation about these access are computed using Lu and O’Connor’s global optimization method.). The differences between the Visual3d standard and Visual3d export to OpenSim angles are further increased because the OpenSim flexion/extension axis and subtalar axis match the orientation specifies in the gait2392_simbody.osim flie. This means the flexion extension axis used to export to OpenSim is skewed compared to the standard Visual3d axis which is a parrellel to the shank X axis and the subtalar axis is really a blend of the standard Visual3d inversion/eversion and toe-in toe-out angles. (If you look in the gait2392_simbody.osim you will see the orientation of these two axes.)

In short when Visual3d is exports to OpenSim it using Lu and O’Connor to compute the rotation about two degrees of freedom specified in gait2392_simbody.osim. When Visual3d computes a regular set of joint angles it first uses Spoor and Velpdaus (1980) to individual compute the position and orientation of each segment (what Lu and O’Connor refer to as segment Optimiztion) and then computes a set of Euler X-Y-Z rotaions to determine flexion-extention, ab/aduction and axial rotation. The two process are entirely different and will not yield the same results.