Fusion: Landscape and Beyond 2.0 | Interactive Art Installation

Summary:

Fusion: Landscape and Beyond 2.0 (2023) is an interactive Artificial Intelligence art installation exploring the memory of AI through the embodiment of landscape texture. Inspired by the traditional brushwork technique Cun, this installation embodies the memory of AI in China by depicting the cultural imprint of Chinese cities and nature with synthetic forms of textural strokes and coding language of AI. Through the use of interactive simulated infrared imagery, the audiences are invited to experience the process of “revealing” and “unrevealing” in the “artificial nature-city,” which is the metaphoric visualization of “infringement” in real life.

Video:

Description:

Fusion: Landscape and Beyond 2.0 (2023) is an interactive Artificial Intelligence art installation exploring the memory of AI through the embodiment of landscape texture. It is a 2.0 version of our previous interdisciplinary art project, Fusion: Landscape and Beyond (2022), which investigates the role of memory in imagination and creation through Artificial Intelligence and Chinese landscape aesthetics. Inspired by the theory of Cultural Memory that emphasizes the process of retrieving and updating memory in responses to the ever-changing situations, we find that the current AI generative image algorithms are analogous to this theory, in that they react to diverse human instructions with previous knowledge. Therefore, we propose a novel concept called AI memory, which can be understood as a kind of institution that manifests the process of preservation and re-embodiment of cultural memory.  

Through this installation, we present a real-time journey of AI memory in China by creating an ever-changing landscape that depicts images of Chinese cities and nature with synthetic forms of textural strokes and coding language of AI. Inspired by the traditional brushwork technique Cun, which is employed by Chinese artists to provide texture to the pictorial elements of nature, we invented a new model of Cun. When it comes to delineating the concrete jungle that we live in today, traditional brushworks that give texture to natural elements are no longer compatible with the city image. And so the new model dynamically fuses AI’s understanding of city aesthetics with traditional Chinese landscape painting brushworks to conceptually depict a synthetic city-nature environment of the past, present, and future.  

We developed a computer vision system that adopts a self-fine-tuned Stable Diffusion model and Clip Interrogator technology to visualize city-nature imagery with the new model of Cun texture in a 3D space. With local APIs, the system can constantly interact with AI models and generate synthetic images of Chinese cities via text-to-image generation. In order to comprehensively explore the memory of AI, the system randomly picks a Chinese city name as input each time. With image-to-image generation, the system then transforms the urbanized world into an artificial nature embedded with Chinese landscape aesthetics. Simultaneously, this system translates the generated visual imprint into AI’s language (prompts) with the clip interrogator algorithm and further be transformed as the texture of the landscape. 

Viewers are invited to walk past the installation as both observers and subjects of observation, engaging with nature as though they are scanning nature with their own bodies. We are interested in the imperceptibility of human infringement on the natural world to those of us who reside in an “artificial nature-city.” For this reason we use simulated infrared imagery to embody humanity’s excessive intrusion into nature and enlighten the unconscious toward ecological balance, creating a cognitive, sensitive, and harmonized way of thinking about the relationship between human and nature. As the simulated imagery of human bodies reflects onto the city image created by AI, it triggers the screen to unveil the underlayer of the landscape image. 

Documentation Images:

Overview
Concept & Inspiration
System Design