Design Efficacy Evaluation of a Landscape Information Modeling–Stable Diffusion (LIM–SD)-based Approach for Ecological Engineered Landscaping Design: A Case Study of an Urban River Wetland

Abstract

This study introduces a Landscape Information Modeling–Stable Diffusion (LIM–SD)-based digital workflow for ecological engineered landscaping (EEL) design, focusing on urban river wetlands. It explores how students from diverse academic backgrounds perform EEL tasks using the LIM–SD approach. A total of 30 participants, including industrial design postgraduates and landscape architecture undergraduates and postgraduates, completed the design tasks. The efficacy of their designs was assessed through expert evaluations on site appropriateness, aesthetics, spatial layout, and eco-engineering techniques of the design proposals, as well as the parametric simulation which calculated the vegetation coverage rate and proportion of riparian areas for each design. Moreover, evaluation of participants' subjective design experiences was conducted via questionnaires. Results indicated that landscape architecture postgraduates outperformed others applying ecological engineering principles. The study also elucidated discrepancies between LIM models and SD-generated renderings, as well as the uncertainty of SD-generated renderings, suggesting improvements are needed to align digital outputs with ecological design criteria.