Author(s)

Chunmei Du ¹, Dahu Lin ¹, Miaomiao Xu ¹, Shuo Sun ¹, Han Zhang ¹

Affiliation(s)

¹ Hebei University of Architecture, Zhangjiakou, China

Corresponding Author

Dahu Lin

ABSTRACT

This paper presents a novel approach to architectural visualization through the development of improved diffusion models capable of local precise redrawing in architectural scheme renderings. We address the critical limitation of current diffusion models in performing targeted, localized modifications while maintaining overall design coherence and architectural accuracy. Our methodology combines advanced diffusion architectures with sophisticated local control mechanisms, including enhanced inpainting techniques, multi-scale attention mechanisms, and architectural domain-specific fine-tuning. Through extensive experimentation on a curated dataset of 10,000+ architectural images, we demonstrate significant improvements in local precision control, achieving a local SSIM score above 0.85 and FID score below 50. Our integrated framework incorporates ControlNet for multi-modal control, LoRA fine-tuning for architectural domain adaptation, and novel loss functions designed specifically for architectural constraints. Human evaluation studies with 15 expert architects and 50 general users validate the practical applicability of our approach, showing professional assessment scores above 4.0/5.0. The proposed system enables architects to perform precise local modifications in seconds rather than hours, fundamentally transforming the iterative design process while maintaining high visual quality and architectural integrity.

KEYWORDS

Diffusion Models, Architectural Rendering, Local Precision Control, Inpainting, ControlNet, Design Automation

CITE THIS PAPER

Chunmei Du, Dahu Lin, Miaomiao Xu, Shuo Sun, Han Zhang, Local Precise Redrawing of Architectural Scheme Renderings Based on Improved Diffusion Models. Journal of Civil Engineering and Urban Planning (2025) Vol. 7: 79-91. DOI: http://dx.doi.org/10.23977/jceup.2025.070212.

REFERENCES

[1] Ho, J., Jain, A., & Abbeel, P. (2020). Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems, 33, 6840-6851.
[2] Zhang, L., et al. (2023). Architectural Rendering with Diffusion Models: A Comprehensive Study. Journal of Architectural Computing, 21(3), 412-431.
[3] Rombach, R., Blattmann, A., Lorenz, D., Esser, P., & Ommer, B. (2022). High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 10684-10695).
[4] Peebles, W., & Xie, S. (2023). Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 4195-4205).
[5] Smith, J., & Johnson, K. (2023). AI-Assisted Architectural Design: Current State and Future Directions. Journal of Design Automation, 45(2), 156-178.
[6] Lugmayr, A., Danelljan, M., Romero, A., Yu, F., Timofte, R., & Van Gool, L. (2022). Repaint: Inpainting using denoising diffusion probabilistic models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 11461-11471).
[7] Xie, S., Zhang, Z., Lin, Z., Hinz, T., & Zhang, K. (2023). SmartBrush: Text and shape guided object inpainting with diffusion model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 22428-22437).
[8] Yu, J., Lin, Z., Yang, J., Shen, X., Lu, X., & Huang, T. S. (2019). Free-form image inpainting with gated convolution. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 4471-4480).
[9] Chen, L., Wang, M., & Liu, H. (2024). Evaluating AI-Generated Architectural Designs: Metrics and Methodologies. International Journal of Architectural Research, 18(1), 89-107.
[10] Zhang, L., & Rao, A. (2023). Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 3836-3847).
[11] Brooks, T., Holynski, A., & Efros, A. A. (2023). InstructPix2Pix: Learning to follow image editing instructions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 18392-18402).
[12] Hu, E. J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., ... & Chen, W. (2021). LoRA: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685.
[13] Anderson, R., & Thompson, S. (2023). Fine-Tuning Diffusion Models for Domain-Specific Applications. Machine Learning Journal, 112(8), 2847-2865.
[14] Liu, X., Park, T., & Wang, Y. (2023). Multi-Scale Attention Mechanisms for Image Generation. Neural Networks, 157, 234-251.
[15] Wang, J., Chen, K., & Zhang, Q. (2024). Boundary-Aware Image Inpainting with Deep Learning. Computer Vision and Image Understanding, 228, 103512.

Sponsors, Associates, and Links

---

• Journal of Sustainable Development and Green Buildings
  
  
• Landscape and Urban Horticulture
  
  
• Bridge and Structural Engineering
  
  
• Soil Mechanics and Geotechnical Engineering
  
  
• Journal of Municipal Engineering
  
  
• Heating, Ventilation and Air Conditioning
  
  
• Indoor Air Quality and Climate
  
  
• Computer Aided Architecture Design