| id |
ecaade2023_66 |
| authors |
Shirahase, Taichi, Fukuda, Tomohiro and Yabuki, Nobuyoshi |
| year |
2023 |
| title |
Developing a Mixed-Reality System with Reflection Rendering of Virtual Objects Using Generative Adversarial Networks |
| source |
Dokonal, W, Hirschberg, U and Wurzer, G (eds.), Digital Design Reconsidered - Proceedings of the 41st Conference on Education and Research in Computer Aided Architectural Design in Europe (eCAADe 2023) - Volume 2, Graz, 20-22 September 2023, pp. 853–862 |
| doi |
https://doi.org/10.52842/conf.ecaade.2023.2.853
|
| summary |
When designing new landscapes, mixed reality (MR) is useful for superimposing virtual models of new buildings on real spaces to visually confirm their relationship to their surrounding environments. However, conventional MR systems are unable to provide accurate results because they cannot represent reflections of virtual models on reflective surfaces. A system has been proposed that uses real-time ray tracing to render reflections of virtual objects on the surface of water, but it does not take into account fluctuations of the water’s surface. To represent reflections, including fluctuations, through advanced calculations such as ray tracing, it is necessary to understand detailed physical wave conditions in real time, which is difficult. In this study, we propose an MR system that can “plausibly” render reflections of virtual objects on the surface of water in real time using a deep-learning model called a generative adversarial network (GAN). We developed a prototype system and verified its quality of reflection rendering and processing speed in a landscape design scene. Our verification results confirmed that using a GAN improves the similarity to the ground truth, and it can be executed at about 6.28 fps. This system enables MR-based waterfront landscape design with reflections on the surface of water, and it contributes to improving consistency for future landscape visualization. |
| keywords |
landscape visualization, mixed reality (MR), generative adversarial network (GAN), reflection, deep learning |
| series |
eCAADe |
| email |
|
| full text |
 file.pdf (3,871,329 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2023/12/10 10:49 |
|
