Diffusion-Based Generative Image Technologies for Urban Cultural Heritage: New Shifts in Public Perception of Authenticity and Cultural Cognition.

Abstract

This paper studies how diffusion-based generative image technology may change public perception of authenticity and cultural cognition in urban cultural heritage interpretation. To avoid viewing generative digital technology as limitless, this paper will focus on text-to-image and image-to-image diffusion systems for generating, modifying and diversifying visual representations of historical streets, buildings and urban areas. This paper presents concepts and practices; it does not conduct original research through data collection of visitors or other empirical evidence to support the conclusions. According to the doctrine of conservation, the theory of authenticity, media theory, digital heritage guidance and technical literature, it is proposed that generative images shift the concept of authenticity from an apparently self-evident visual attribute to a judgment about evidence, provenance, uncertainty and institutional trust. They provide comparisons and questions, but photorealism may also conceal speculation and reduces controversial urban history. Therefore, the responsible workflow should separate records from hypotheses, disclose the generation process, keep the source link, mention communities, and not present them as fact by virtue of visual plausibility.