The Concretization of Our Imagery

By Kaiyu Cheng

Supported by the University of Chinese Academy of Sciences Education Foundation

Installation & Interactive Art

Drawing inspiration from “mood” and “atmosphere” and translating them into installations that drive formal innovation has proven a practical approach to breaking free from the constraints of traditional forms. Architects and artists, through acute social observation and rich modeling experience, have successfully enriched our urban spaces. However, excessive reliance on subjective judgments by elite individuals is not only limited by personal capabilities but also contradicts the trend of public participation. Today, AIGC trained on internet big data combine comprehensive information processing with intuitive interaction, offering opportunities for non-professional (or cross-disciplinary) groups to engage in urban renewal.

Installation The Concretization of Our Imagery comprehensively employs AIGC technologies including NLP models (ChatGPT 4.0, DeepSeek-R1), Text- image Translation Models (Midjourney, Stable Diffusion), and Metric Depth Estimation Models (ZoeDepth). Based on extensive collection and analysis of local residents’ urban imagery (through linguistic and visual media), it translates abstract concepts into 2D and 3D spaces, and physical realization through 3D printing.

Modern AI programs empower non-professionals with visualization capabilities, while big data-translated imagery installations provide designers and the public with their first shared visual reference: Public imagery can now be expressed through installations without requiring professional training, while decision-makers are compelled to understand and develop these installations to address public concerns. Throughout this process, the spirit of public participation manifests in both imagery collection and design communication, while innovative design methods enrich both conceptual expression and formal breakthroughs.

This work is supported by: Beijing Urban Governance Research Base (2024CSZL01); Education Foundation of the University of Chinese Academy of Sciences (F100875).