Embodied Animation

Description:

Embodied Animation explores animation as an embodied, performative, and spatial practice. Rather than producing moving images through frame-by-frame drawing or pre-rendered computational pipelines, the project generates animation through the real-time movements of the human body. Hand gestures are captured as continuous spatial and temporal data and translated through generative AI into evolving visual forms, transforming the body from a controller of animation into its primary performative medium.

The work originates from a live interactive system in which the performer’s hand functions as a “living brush.” Gestures continuously shape image, motion, texture, and sound through a feedback loop between bodily movement and machine generation. Instead of determining every visual outcome in advance, the performer improvises with the generative system, allowing images to emerge through the negotiation between human intention, physical gesture, and algorithmic interpretation. Animation therefore becomes not simply a sequence of images, but the trace of an embodied event.

Conceptually, Embodied Animation draws from performative image-making traditions such as hand-shadow puppetry, sand painting, and gestural drawing, practices in which movement, temporality, and ephemerality are inseparable from the image itself. By connecting these embodied traditions with contemporary generative AI, the work investigates how bodily intelligence can remain central within increasingly automated systems of image production. Here, the machine does not replace the gesture; it amplifies, interprets, and transforms it.

Through this transformation, Embodied Animation proposes an expanded understanding of animation: not as imagery authored entirely in advance, but as a record of embodied negotiation between human and machine. Projected onto the city, these ephemeral gestures become collective-scale images, extending the performing body beyond its physical boundaries and turning architecture itself into a surface for embodied motion.

Technical Information:

For its urban projection presentation, the live generative performances are recorded and reconfigured as large-scale moving images projected onto architectural façades. The building becomes a new spatial body for the animation: gestures originally produced at the scale of the hand expand across windows, surfaces, edges, and volumes of the city. This shift from intimate bodily movement to monumental architectural projection creates a dialogue between body, image, machine, and urban space.

Technically, the original performance system integrates Leap Motion hand tracking, TouchDesigner, TouchOSC, a LoRA-trained diffusion model, and StreamDiffusion to enable real-time visual generation. The resulting animation preserves the temporal continuity and unpredictability of the live interaction while translating it into a cinematic and architectural format.

Additional Information:

Exhibition Site: Nanfeng City Screen 2