Journal of Digital Landscape Architecture

Seite drucken

Accuracy Evaluation of Tree Images Created Using Generative Artificial Intelligence

This study evaluated the precision of artificial intelligence (AI) in generating tree images and aimed to integrate AI advancements with landscape design needs, fostering more innovative and efficient design approaches. The AI image-generation tool, Midjourney, was used to assess the accuracy of images for three tree species, to determine its applicability to landscape design. The results demonstrate that although the AI emphasized the color characteristics and seasonal changes of the trees with a certain level of accuracy, it exhibited limitations in accurately replicating the overall form and texture of the trees, particularly the complex patterns and textures of the bark. Consequently, AI image-generation technology still faces challenges in accurately reproducing the detailed characteristics of natural elements, which underscores the enhancements needed for practical applications in landscape design. These findings suggest AI’s potential as a rapid visualization tool in the initial stages of landscape design while highlighting the need for more sophisticated technological development and improvements.

Autor / Author: Kim, Jaeheon; Lee, Yumi
Institution / Institution: Seoul National University, Seoul/Republic of Korea; Seoul National University, Seoul/Republic of Korea
Seitenzahl / Pages: 9
Sprache / Language: Englisch
Veröffentlichung / Publication: JoDLA – Journal of Digital Landscape Architecture, 9-2024
Tagung / Conference: Digital Landscape Architecture 2024 – New Trajectories in Computational Urban Landscapes and Ecology
Veranstaltungsort, -datum / Venue, Date: Vienna University of Technology, Austria 05-06-24 - 07-06-24
Schlüsselwörter (de):
Keywords (en): Text-to-image AI, image generation, design tool, tree representation, image accuracy
Paper review type: Full Paper Review
DOI: doi:10.14627/537752098
8041 -