Urban Architect: Steerable 3D Urban Scene Generation with Layout Prior
https://github.com/UrbanArchitect/UrbanArchitect/assets/13656744/ffd06914-f522-44a2-9a44-93a516506e36
Introduction
The repository contains the official implementation of source code and pre-trained models of our paper:"[Urban Architect: Steerable 3D Urban Scene Generation with Layout Prior]()". It is a novel pipeline for large-scale 3D urban scene generation!
Updates
- 2024.04.26: We release samples of layout in
dataset/data
and pretrained ControlNet weights!
- 2024.04.11: The :fire::fire::fire: pre-print :fire::fire::fire: is released! Refer to it for more details!
- 2024.04.10: The project page is created. Check it out for an overview of our work!
Environments
- Main requirements:
- Download the pretrained model weights of Stable Diffusion
- Download the pretrained model weights of CLIP
- Download the pretrained model weights of ControlNet
- Other requirements are provided in
requirements.txt
Generate a Scene
Coarse stage
Please refer to train.sh
.
Refinement stage
Please refer to refine.sh
.
Render video
Please refer to render.sh
.
To-Do List
- [ ] Release 3D layout data
- [x] Technical Report
- [x] Project page
Related Work
- (ICCV 2023) Urban Radiance Field Representation with Deformable Neural Mesh Primitives, Fan Lu et al. [Paper], [Project Page]
Citation
If you find this project useful for your work, please consider citing:
@article{lu2024urban,
title={Urban Architect: Steerable 3D Urban Scene Generation with Layout Prior},
author={Lu, Fan and Lin, Kwan-Yee and Xu, Yan and Li, Hongsheng and Chen, Guang and Jiang, Changjun},
journal={arXiv preprint arXiv:2404.06780},
year={2024}
}