Carnegie Mellon University | Feb 2023 - May 2023 | Course Project, Team of Three
We Introduced a novel approach to architectural stylization using NeRF-based 3D building representations, which can be used to generate highly detailed and realistic stylized scenes.
We proposed several approaches to induce geometry warping during the architectural style optimization process while preserving the background appearance, including patch-based NNFM loss and auxiliary CLIP loss from text descriptions.
We demonstrated the effectiveness of our approach to accurately capture the desired stylization effects, through experimental results.