CS677-Parallel-optimised-View-Synthesis

Setup for our code

1. Create a conda environment using our environment.yml file.

   
   conda env create -f environment.yml
   
   conda activate paraenv 
   

2. Create viewpoints using make_data.py with N_VIEWS viewpoints (stored in data.csv)

   python3 make_data.py N_VIEWS
   

3. Run the MPI code surf_optim for surface rendering for NUM_PROCESSES processes and N_VIEWS viewpoints (where N_VIEWS > NUM_PROCESSES and number of rendering processes is NUM_PROCESSES - 1)

    mpiexec -n NUM_PROCESSES pvbatch surf_optim.py N_VIEWS
   

   Run the code vol_optim for volume rendering for NUM_PROCESSES processes and N_VIEWS viewpoints (where N_VIEWS > NUM_PROCESSES and number of rendering processes is NUM_PROCESSES - 1)

    mpiexec -n NUM_PROCESSES pvbatch vol_optim.py N_VIEWS
   

   The initial images will be saved in InitImages and Optimised images will be saved in OptiImages

4. The time taken for surface rendering optimization is stored in results1.csv and volume rendering optimization is stored in results2.csv. The columns in results1/2.csv represent number of processes, number of viewpoints, time taken.

5. This can be analysed by opttime.

6. Similarity results can be reproduced using similaritygraphs.ipynb. The similarity and opttime codes are in the helper files.