SceneDreamer360: Text-Driven 3D-Consistent Scene Generation with Panoramic Gaussian Splatting

Wenrui Li, Fucheng Cai, Yapeng Mi, Zhe Yang, Wangmeng Zuo, Xingtao Wang, Xiaopeng Fan

[🌐 Project Page] [📜 Paper]

Introduction

SceneDreamer360 leverages a text-driven panoramic image generation model as a prior for 3D scene generation and employs 3D Gaussian Splatting (3DGS) to ensure consistency across multi-view panoramic images. Specifically, SceneDreamer360 enhances the fine-tuned Panfusion generator with a three-stage panoramic enhancement, enabling the generation of high-resolution, detail-rich panoramic images. During the 3D scene construction, a novel point cloud fusion initialization method is used, producing higher quality and spatially consistent point clouds. Our extensive experiments demonstrate that compared to other methods, SceneDreamer360 with its panoramic image generation and 3DGS can produce higher quality, spatially consistent, and visually appealing 3D scenes from any text prompt.

Visualization

Environment Setup

Follow these steps to set up the required environment:

conda env create -f environment_strict.yaml
conda activate scenedreamer360

pip install -r Enhance_img/requirements.txt

cd PanoSpaceDreamer/submodules/depth-diff-gaussian-rasterization-min
python setup.py install
cd ../simple-knn
python setup.py install
cd ../../..

Checkpoints

Download PanFusion checkpoints and move to logs/4142dlo4/checkpoints. Also, download the panoramic optimization model from Baidu Cloud, and extract models.zip into Enhance_img/.

Running

Modify the parameters in the config.json file to suit your test cases.

Update the text field to point to the prompt file you wish to test. campath_gen(default: fullscan) supports the following options: fullscan, layerscan, lowerscan, rotate360, lookaround, moveright, moveback, arc, lockdown, hemisphere. campath_render(default: 1440) supports the following options: 1440, 360, rotate360, headbanging, llff, back_and_forth. To run the test, use this command:

WANDB_MODE=offline WANDB_RUN_ID=4142dlo4 python main.py predict --data=Matterport3D --model=PanFusion --ckpt_path=last

Batch Processing

For batch testing, list the test prompts in the data/prompt.txt file (one prompt per line), and update the file name in the data/Matterport3D/mp3d_skybox/e9zR4mvMWw7/blip3_stitched directory to test.txt.

Run the following command to start batch processing:

python run.py

Results

The results of the runs will be saved in the logs/4142dlo4/predict directory.