| id |
sigradi2024_27 |
| authors |
Abdallah, Yomna K., Estevez, Alberto T., Lu, Sheau T., Almaraz, Julia, Cuellar Loor, María Del Carmen, Mendoza Estrada, Jaren, Lagos Suarez, Juan A., Melachropoulou, Konstantina and Pacheco Silva, Sandra N. |
| year |
2024 |
| title |
3D Printed Biodigital Fractal Bioreceptive Topologies from Diffusion Models And 2D CNNs Generated Design Through 2.5D Depth Mapping |
| source |
Herrera, Pablo C., Gómez, Paula, Estevez, Alberto T., Torreblanca-Díaz, David A. Biodigital Intelligent Systems - Proceedings of the XXVIII Conference of the Iberoamerican Society of Digital Graphics (SIGraDi 2024) - ISBN 978-9915-9635-2-5, iBAG-UIC Barcelona, Spain, 13-15 November 2024, pp. 827–840 |
| summary |
AI-Aided Design (AIAD) tools have revolutionized the design process by rapid customized high-resolution renders. Yet, they were limited in their 3D-translation and direct fabrication. Currently, AI-models for 3D-depth-mapping are evolving and they require a criteria for their integration in the design process to maintain the human authorship and creativity to achieve sustainability. The current work reports an (AIAD) to fabrication study of Bio-receptive tiles for integration in the built environment. The experimental design methodology includes (AIAD) phases of prompt synthesis, data generation and augmentation. Employing AI-Diffusion models, Convolution Neural Networks, Recurrent Neural Networks, and transformer models. Followed by 2D to 3D-depth-mapping from a single 2D-image and 3D-printing into bio-receptive tiles. The 2D-CNN image-to-sequential data generation proved to be an efficient generative design tool with more control over image-generation operative parameters than Diffusion models. The 3D-printed bio-receptive tiles are time-material-cost sustainable with high resolution and multi-scale topologies for hosting microbial strains. |
| keywords |
Bioreceptive Topologies, Convolution Neural Networks, AI-Aided design, Depth-mapping, Recurrent Neural Networks, LSTM |
| series |
SIGraDi |
| email |
|
| full text |
 file.pdf (2,732,715 bytes) |
| references |
Content-type: text/plain
|
| last changed |
2025/07/21 11:48 |
|
