Motion Sculpture (In-progress)

The project explores how to visualizes motion capture data as sculptures. The body motion is tracked by human body joints. A human skeleton is represented by 22 joints. Each of them draw a 3D curve in the air during the motion. The most dramatic 3D curves come from the joints on the four limbs, each of which have four joints.

The initial idea is to create four curved surfaces, one for each limb, by connecting the four 3D curves shaped by the motion trajectories of the four joints on each limb.

Each motion data sample is a matrix with a dimension of 240 x 66. It consists of 240 frames of motion, each motion recording the the XYZ coordinate values of the 22 body joints. The recording was 60 fps, which means the motion lasts 4 seconds.

The data is stored as csv files. The csv file is parsed by Grasshopper as a data tree with 240 branches, each of which has 66 numbers. Each set of 66 values can be interpreted as 22 points by Num2Pt component. With a flip modules to convert the 240 sets of 22 points into 22 sets of 240 points. The following polyline module makes the 22 sets of 240 points into 22 polylines.

Using the Split Tree module can pick out cures of selected body joints. Here I creates four groups of polylines, each for one limb, by specifying the joint index numbers of each limb. To create surfaces that go across the four curves of each limb, I used loft function.

Simultaneously displaying the four surfaces generated by four limbs produces the complex sculptural object. However, it might be difficult to print out. I decided to choose one of them to print. The second image shows how they look after take them apart.

The selected surface is extruded for 1mm to create a volumetric object. It is exported to Cura and the previewer complained about the surfaces. Apparently, it is challenging to print unless there is no gravity.

The model is being printed. We will see how it looks.

----------

It was printed out last night. It does not look great. I will modify the modeling methods.