Text2Light主要目标是生成室内和室外的全景图,没有人。生成的图片光线比较’暗,T4上每次生成需要3分钟以上。
1.安装conda
!wget https://repo.anaconda.com/archive/Anaconda3-2022.10-Linux-x86_64.sh && bash Anaconda3-2022.10-Linux-x86_64.sh
2.创建并激活虚拟环境
!/root/anaconda3/bin/conda env create -f /content/Text2Light/environment.yml
!/root/anaconda3/bin/conda activate text2light #若base没有激活会失败
3. 激活base
!conda init bash
!source activate
4. 进入虚拟环境
#若环境安装失败
!sudo apt-get install libomp-dev
!pip install ftfy regex tqdm omegaconf pytorch-lightning tensorboardX einops transformers
!pip install kornia
!pip install imageio-ffmpeg
!pip install faiss
!pip install opencv-python==4.1.2.30
5. 若cv2启动失败
!apt-get update
!apt-get install libsm6
!apt-get install libxrender1
!apt-get install libxext-dev
6. 终于可以执行命令了
!python text2light.py -h
7.下载预训练模型到logs目录下(https://drive.google.com/drive/folders/1HKBjC7oQOzrkGFKMQmSh6PySv6AycDS3?usp=sharinghttps://drive.google.com/drive/folders/1HKBjC7oQOzrkGFKMQmSh6PySv6AycDS3?usp=sharing)
8. 可以愉快的生成了
python text2light.py -rg logs/global_sampler_clip -rl logs/local_sampler_outdoor --outdir ./generated_panorama --text "YOUR SCENE DESCRIPTION" --clip clip_emb.npy --sritmo ./logs/sritmo.pth --sr_factor 4