IMPLEMENTASI PENGGUNA ARTIFICIAL INTELEGENT (AI) PADA PLATFORM MIDJOURNEY DALAM ITERASI ARSITEKTUR

Abstract

The field of architecture has experienced spectacular developments over the years. The reason is that in the Industrial 4.0 era, various breakthroughs and disruptions to the world of architecture have resulted in rapid developments that enable the design process to create more unique and more complex designs. AI (Artificial Intelligence) technology which is currently developing in various scientific fields has the potential for future architectural developments. One of the popular AI developments is the ability to convert images into text which provides potential in the architectural iteration process or what is better known as the phase of looking for reference ideas in the schematic stage. The process of converting certain text or keywords into images can actually help architects to find patterns and shapes that suit preferences, speeding up the design process and producing efficient designs. In addition, AI can predict design performance in various scenarios. It is an attempt to use artificial intelligence to explore early architectural design concepts. The data analysis method used in this research was to observe the effect of using keywords in creating prompts or commands for AI on Midjourney and then comparing them with the results of other images. After the comparative study, it will be measured to what extent the use of AI is implemented in certain situations or conditions in the Architectural Iteration phase that occurs. Based on the results of image analysis, the prompt arrangement is also a control variable used to create specific drawings in creating architectural concept iterations consisting of "Object Type, Material Type, Architectural Style, Architect Name, Style Render, View, Environment/Background and Type of Lighting" With the results of the prompt translation, it can be concluded that the more specific the order of the prompts used as keywords, the more specific the resulting image will be, in this case we can exercise control over the image that we want to generate.