> Motivation
> Example Movie
> Comment
> Conclusion
> Related Experiments
> Setup

 


Performance Photo

Screenshot

> Top

 

Motivation

Like this experiment began with the question if we perceive a difference in the representation of a previously captured and just replayed motion and a motion which is captured in real-time. We chose to represent a sensor by a box and the motion path of the sensor by a line and to visualy overlay the recorded and the live captured sensor data. Conceiving this representation, the question occured in how far motion capture could contribute to dance notation.

Dance notation is a form of writing down movements with most of its spatial and temporal details, so that the dance can be taught from this recorded notation at a later time. Nowadays classical dance notation forms like Laban or Benesh are little in use. Instead most dance companies use video-recordings to archive and later re-teach choreography. In how far can we utilize motion capture for notating some information of a movement ? In our case: which part of the information is kept in the captured movement trace, which information gets lost ?

 

> Top

 

Example Movie

Stage View
Computer Image
Two movement sequences are repeated several times. Each time the movement is captured via a sensor attached to one hand and is represented by a thin tube in space. The faster the motion, the thicker the tube. The captured movement traces are repeatedly played back as a recording and overlay with the live-captured data.

 

> Top

 

Comment

This experiment touches the field of dance-notation and movement analysis. The movements Nik chooses are rather simple: a turn, a jump, swinging of the arms. While moving he silently counts a rhythm in order to stay as precise with the timing as during the previous repetitions. Nik performs the repetitive movement and by pressing the record button (the RF button he holds in his hand) he can capture and leave a kind of rhythmic history of the movement he has just done. One can see in how far the recorded movement line of a previous motion matches the current live motion and its line. How accurate can a movement be repeated, not only in its spatial but also in its temporal exactness?

In the video-example one can compare the two sequences that Nik moves through and leaves as recordings on the screen. In the first sequence the two movement lines and also the two cubes that indicate the sensors' position (left hand) match in paths and timing. In the second sequence the repeating of the timing is not as precise. Especially when it comes to dynamics and timing, this might be a helpful tool to document a dance. Less in dancing (where the learning of an existing choreography is also a question of individual interpretation) but more in sports like swimming, a movement analysis as it is done in this experiment might be valuable. While going through a repetitive motion like one does in swimming, an ideal motion-path and -timing can be compared with the athlet's actual path and timing.

In most experiments we decided to reduce the amount of sensors to one or two. By doing this, the focus of a movement that is performed by the entire body is directed to the one or two limbs where the sensors are attached. This experiment leads the viewer's focus to the arm path and timing, the phrasing of the movements. It is rare that a body movement gets isolated and fragmented in this way, that some parts become more visible than others. As a viewer one is left with this one fragment on the screen even when the dancer has stopped moving. How much of the original body movement do I remember just by viewing this one fragment?

To evaluate the suitability of this experiment and as a dance notation tool, we tested in how far one can repeat the original motion given just the computer generated motion picture. As expected, it turned out that the movement of the hand(s) was the most accurate, while the movement of the rest of the body was driven by the support of the hands' movements.

> Top

 

Conclusion

The initial question, in which respect a recorded/ played-back motion and a motion that is captured in real-time differ, is answered in the simplest way by a comparison with a singer on stage. Imagine a singer imitating the singing while the song is played back from tape. It obviously lacks the sense of the authenticity of the live performance. In case the singer sings live and is just additionally supported by a prerecorded voice, the authenticity is still there. Overlaying both, in our setup there was no fundamental opposition between the recorded and the live data, it added just a level of complexity.

It seems worthwhile to follow up evaluating the suitability of motion capture as a dance notation tool by adding additional sensors or finding alternative representations. As opposed to software which shows a literal representation of the complete body like Dance Forms (former Life Forms), using such an abstract but temporal and spatially accurate representation gives a dancer more space for interpretations. On the other hand such a representation has to be supported by a descriptive notation of the motivation and concept of the movement.

 

> Top

 

Related Experiments

 

> Top

 

Setup


 

Hardware:

  • Graphics PC
  • Projector
  • Polhemus tracking system: 1 Sensor
  • RF Button



Configuration:

  • The Sensor is attached to the dancer's left hand. The position and orientation of the sensor is controlling the position and orientation of a small cube. When moved, the cube leaves a trace in the form of a narrow tube. The tube's thickness is related to the speed of the cube's motion. The faster the cube is moved, the thicker the diameter of the tube. The Technical Screenshots below show the sensor as a green ball at the origin of the cube.
  • The RF Button is attached to the dancer's right hand. A button press starts the recording of the data while the cube with its trace is still controlled by the sensor. The next button press finishes the recording. A new additional cube with a trace appears on the screen, which is controlled by the looped playback of the recorded data. A maximum of five of these recordings are played back in parallel. Each further recording overwrites the oldest one. With these button presses, the dancer determines the beginning and end of a motion unit.

 


Technical Setup

Spatial Setup

Technical Screenshot 1

Technical Screenshot 2
> Top