Environment Setup
- We provide a conda env file that contains all the required dependencies
conda env create -f environment.yml - Following this, you can activate the conda environment with the command below.
conda activate mirrordiffusion
Download Dataset
- C2D-F and C2G-F: [Google Drive].
- H2Z-F: [Google Drive].
- S2O-F: [Google Drive].
Real Image Translation
- First, run the inversion command below to obtain the input noise that reconstructs the image.
python src/null_inversion.py --input_image ./assets/test_images/cat/ --results_folder ./exp_output/stable_cat/ --num_ddim_steps 60
- Next, we can perform image editing with the editing direction as shown below.
python src/null_edit.py --inversion "test/stable_cat/null_inversion/" --prompt "test/stable_cat/prompt/" --task_name "cat2dog" --results_folder ./exp_output/sketch_cat/ --num_ddim_steps 60
Finding Custom Edit Directions
python generate_direction.py
This project is build based on Pix2Pix-Zero and Null-Text Inversion. We thank the authors for sharing their code.