Collective Improvisation I (CiI)
Interactive experience (installation)
Human beings are living in a divided world. A time of polarisation. From profound rooted issues and conflicts relating to racism to recent events like the pandemic. This research project involves easing the hostilities between people with different backgrounds (cultural, racial, religious, et cetera). Using machine learning and an interactive dance installation as a medium to bring people closer together, find a sense of connection with each other and reduce hostility and prejudice.
On the streets, everyone puts their guards up. Within a two-meter-wide pavement, people tend to walk as far apart as they can when confronted with other strangers. We are so closed and guarded into our comfort zones, even afraid to do anything embarrassing, but when we are dancing, we seem to let go of some of those feelings and unconsciously put our guards down, especially on the dance floor. People do not mind other strangers dancing within only half a meter. The connectedness that dancing creates intrigues the author; he wants to understand why it makes us open up like that.
The result is an interactive installtion (prototype). It invites people to follow, a Machine Learning ‘choreography’ moving image and dance either alone or with other audience, to experiment if the machine learning result can enhance the sense of connnectedness to each other. The installation will record the dance and automatically put them on the right side of the screen, creating a Asynchronous connection between the different audiences as well as for Zihao’s further research.
Collective Improvisation I
Length: 2 min 09 sec
On the other hand, animals do not have or have fewer biases and discriminations. Zihao brought in animal and territoriality study to explore whether people can learn something from how they cope/deal with multi-species and trans territories by observing their body language and dancing behaviour.
Zihao collected photographs and videos of human and non-human dancing patterns and movements. He used his body as a unified model to mimic the dance moves, creating images that form datasets. These datasets form the 'knowledge' to teach and feed Zihao's AI machine model. The aim is to find hidden links and dance choreographies that humans would have never imagined.
These new choreographies are shown to Zihao's audience to test whether they can stimulate the urge to dance and bring people closer together to feel connected.
dancing movement collection, dataset 02, in total 1670 pics
Machine Learning Result 02
Machine Learning Result 02 - ‘Latent Walk’
Even though, it doesn’t look like dancing at all, this is a crucial experiment for the project. It made it clear that A.I. ‘sees’ dancing movement as a pure database - colour, pixel pieces, and statistics; although these might not result in the final outcomes of the project, the process contributed to the logic.
In order to 'teach' it the right ‘knowledge’ and try to make it ‘dance,’ instead of figuring the visual moves, Zihao introduced vector and math coordinates into the following experiments.
Converted Coordinates
Coordinate collection, dataset 03, in total 1670 pics
Machine Learning Result 03 - ‘Mural Dance’