- 會議論文
從三維草圖到概念設計－AI生成技術於室內設計的應用
From 3D Modeling to Conceptual Design: Application of AI Generation Technology to Interior Design
摘要
自2022年開始，生成式AI的成長影響全球出版、設計等產業之工作流程，空間設計的模擬也是生成式AI發展的強項，因此，本研究將室內設計常見之粗略三維模擬影像，透過原圖模擬與關鍵字，以現今主流生成式AI程式重新計算，包含Midjourney與LOOKX，進行兩者影像生成效果之比較，運算的結果由室內設計學系碩士生與從業人員，依據與原圖之還原度、藝術表現等因子進行問卷調查，並從結果中得知Midjourney具有豐富的藝術表現，但與原始影像之契合度不佳，LOOKX計算的結果與原圖相似度高，但影像的變化度則較少。兩者均可作為不同階段之設計輔助使用。建議未來的研究方向，隨著各類生成式AI的進步，可增加不同AI工具，以及不同類型之室內設計圖像生成比較。用以進行更嚴謹之可行性評估。
並列摘要
Since 2022, the growth of generative AI will affect the workflow of global publishing and design industries, and the simulation of spatial design is also a strong point in the development of generative AI. Therefore, in this study, we compare the image generation effect of the common rough 3D simulated images of interior design through the simulation of the original drawings and the keywords, and re-calculate the images with the mainstream generative AI programs, including Midjourney and LOOKX, and compare the results of the two. The results of the calculation were obtained from a questionnaire survey conducted by master students of the Department of Interior Design and practitioners based on factors such as the degree of restoration and artistic expression of the original image. From the results, it was found that Midjourney has rich artistic expression but does not match well with the original image, while the results of the calculation of LOOKX have a high degree of resemblance to the original image but with less variation in the image. Both of them can be used in different stages of design. It is suggested that the future research direction, with the progress of various types of generative AI, can increase the different AI tools, as well as different types of indoor design image generation comparison. This will be used for more rigorous feasibility assessment.