Journal of KIBIM (한국BIM학회 논문집)

Korean Institute of Building Information Modeling (한국BIM학회)
권호 표지
DOI QR코드

DOI QR Code

Generative AI-based Exterior Building Design Visualization Approach in the Early Design Stage - Leveraging Architects' Style-trained Models -

생성형 AI 기반 초기설계단계 외관디자인 시각화 접근방안 - 건축가 스타일 추가학습 모델 활용을 바탕으로 -

  • Yoo, Youngjin ;
  • Lee, Jin-Kook
  • 유영진 (연세대학교 실내건축학과) ;
  • 이진국 (연세대학교 실내건축학과)
  • Received : 2024.06.17
  • Accepted : 2024.06.20
  • Published : 2024.06.30
Download PDF
⟨ Previous Next ⟩

Abstract

This research suggests a novel visualization approach utilizing Generative AI to render photorealistic architectural alternatives images in the early design phase. Photorealistic rendering intuitively describes alternatives and facilitates clear communication between stakeholders. Nevertheless, the conventional rendering process, utilizing 3D modelling and rendering engines, demands sophisticate model and processing time. In this context, the paper suggests a rendering approach employing the text-to-image method aimed at generating a broader range of intuitive and relevant reference images. Additionally, it employs an Text-to-Image method focused on producing a diverse array of alternatives reflecting architects' styles when visualizing the exteriors of residential buildings from the mass model images. To achieve this, fine-tuning for architects' styles was conducted using the Low-Rank Adaptation (LoRA) method. This approach, supported by fine-tuned models, allows not only single style-applied alternatives, but also the fusion of two or more styles to generate new alternatives. Using the proposed approach, we generated more than 15,000 meaningful images, with each image taking only about 5 seconds to produce. This demonstrates that the Generative AI-based visualization approach significantly reduces the labour and time required in conventional visualization processes, holding significant potential for transforming abstract ideas into tangible images, even in the early stages of design.

Keywords

Acknowledgement

본 연구는 국토교통부/국토교통과학기술진흥원의 2024년도 지원으로 수행되었음(과제번호 : RS-2021-KA163269). 본 연구는 미래융합연구원(ICONS), 연세대학교 지원으로 수행되었음(과제번호 : 2019-22-0043).

References

  1. Azhar, S. (2011). Building Information Modeling (BIM): Trends, Benefits, Risks, and Challenges for the AEC Industry, Leadership and Management in Engineering, 11(3), pp. 242-252. 
  2. Betker, J., Goh, G., Jing, L., Brooks, T., Wang, J., Li, L., Ouyang, L., Zhuang, J., Lee, J., Guo, Y., Manassra, W., Dhariwal, P., Chu, C., Jiao, Y., Ramesh, A. (2023). Improving Image Generation with Better Captions, arXiv preprint arXiv: 2006.11807. 
  3. Chan, C. S. (1992). Exploring Individual Style in Design, Environment and Planning B: Planning and Design, 19(5), pp. 503-523. 
  4. Chan, C. S. (1994). Operational Definitions of Style, Environment and Planning B: Planning and Design, 21, pp. 223-246. 
  5. Cheon, S., Kim, S., Lee, S. (2022). A Study on the Potential of Development of Korean Architecture Using AI - New Typology Suggestion Using StyleGAN -, Proceedings of Spring Annual Conference of Architectural Institute of Korea, 42(1), pp. 197-200. 
  6. Ching, F. (2023). Architecture - Form, Space, and Order -, John Wiley & Sons, Inc. 
  7. Chiu, M. L. (1995). Collaborative Design in CAAD Studios: Shared Ideas, Resources, and Representations, Proceedings of International Conference on CAAD Future, pp. 749-759. 
  8. Cho, D., Lee, J. K. (2021). Training Floorplan Sketches and Applying to the Spatial Design - Focused on the Development of Automated BIM Modeling module from Floor Plan Sketches in the Early Stage of Design, Journal of Korea Intitute of Spatial Design, 16(3), pp. 365-374. 
  9. Choi, J., Lee, J. (2022). A Study of Interior Style Transformation with GAN Model, Journal of Korean Institute of Building Information Modeling (KIBIM), 12(1), pp. 55-61. 
  10. Choi, S., Lee, J. K. (2023). Generative AI for Automated Urban Housing Floor Plan Generation, Korean Journal of Computational Design and Engineering, 28(4), pp. 514-523. 
  11. Eastman, C. (2009). Automated Assessment of Early Concept Designs, Architectural Design, 79(2), pp. 52-57. 
  12. Enscape (2015). https://enscape3d.com (Feb. 15. 2024). 
  13. Evolvelab (2023). Veras, https://www.evolvelab.io/veras (Feb. 15. 2024). 
  14. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y. (2014). Generative Adversarial Nets, Advances in Neural Information Processing Systems, 27. 
  15. Greenberg, D. P. (1974). Computer Graphics in Architecture. Scientific American, 230(5), pp. 98-107. 
  16. Han, S., Shin, D. (2021). A Study on Architectural Image Generation using Artificial Intelligence Algorithm - A Fundamental Study on the Generation of Due Diligence Images Based on Architectural Sketch -, Journal of Korean Institute of Building Information Modeling (KIBIM), 11(2), pp. 54-59. 
  17. Hartoonian, G. (2016). Materiality Matters - If only for the Look of it! In Loschke, S.K. (Ed.), Materiality and Architecture, Routledge, pp. 59-78. 
  18. Hessel, J., Holtzman, A., Forbes, M., Bras, R. L., Choi, Y. (2022). Clipscore: A Reference-free Evaluation Metric for Image Captioning. arXiv:2104.08718. 
  19. Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W. (2021). Lora: Low-rank Adaptation of Large Language Models. arXiv preprint arXiv:2106. 09685. 
  20. Jeong, H., Lee, J. K. (2023). Image Gen AI-based Model Fine-Tuning and Design Application, Korean Journal of Computational Design and Engineering, 28(4), pp. 524-536. 
  21. Jo, H., Lee, J. K., Lee, Y. C., Choo, S. (2024). Generative Artificial Intelligence and Building Design: Early Photorealistic Render Visualization of Fac̡ades Using Local Identity-trained Models, Journal of Computational Design and Engineering, 11(2), pp. 85-105. 
  22. Lee, K. A. (2018). A Research on Space Identity Developing model by Design Process, Journal of Digital Convergence, 16(12), pp. 563-569. 
  23. Lee, W., Kim, S., Yu, Y., Koo, B. (2022). Development of Graph Based Deep Learning methods for Enhancing the Semantic Integrity of Spaces in BIM Models, Korean Journal of Construction Engineering and Management, 23(3), pp. 45-55. 
  24. Lumion (2010). https://lumion.com (Feb. 15. 2024). 
  25. Moussavi, F. (2021). The Function of Style. Actar D, Inc. 
  26. Oppenlaender, J. (2022). The Creativity of Text-to-image Generation, Proceedings of the 25th International Academic Mindtrek Conference, pp. 192-202. 
  27. Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., Chen, M. (2022). Hierarchical Text-Conditional Image Generation with CLIP Latents. arXiv preprint arXiv:2204.06125. 
  28. Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B. (2022). High-resolution Image Synthesis with Latent Diffusion Models, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 10684-10695. 
  29. Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E., Kamyar, S., Ghasemipour, S., Karagol, B., Mahdavi, S., Gontijo, R., Salimans, T., Ho, J., Fleet, D., Norouzi, M. (2022). Photorealistic Text-to-image Diffusion Models with Deep Language Understanding, Advances in Neural Information Processing Systems, 35, pp. 36479-36494. 
  30. Sandaker, B. N., Kleven, B., Wang, A. R. (2022). Structural Typologies and the Architectural Space - Studies of the Relationship Between Structure and Space by Application of Structural Types to Multistory Buildings. Architecture, Structures and Construction, 2, pp. 199-221. 
  31. Schapiro, M. (1961). Style. Aesthetics Today, pp. 81-113. 
  32. Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., Ganguli, S. (2015). Deep Unsupervised Learning Using Nonequilibrium Thermodynamics. ICML, pp. 2256-2265. 
  33. Twinmotion (2005). https://www.twinmotion.com (Feb. 15. 2024) 
  34. Yoo, Y., Lee, J. K. (2023). Generative AI-based Construction of Architect's Style-trained Models and its Application for Visualization of Residential Houses, Design Convertgence Study, 22(6), pp. 103-116. 
  35. Zhang, L., Agrawala, M. (2023). Adding Conditional Control to Text-to-image Diffusion Models, arXiv preprint arXiv: 2302.05543.