We are tasked on evaluating vvenc's performance based on a different number of threads and a few settings (mainly whether we use 2x2 row x column tiles and whether WaveFront synchronization is on or off). These are scripts that help generate data for an easier analysis of the codec.
Not an analysis (should have changed the title...)
Feel free to contribute anything after the deadline has passed (1st of March 2024).
This is a joint effort by Dimosthenis Krallis (wallmenis) and David Parathyras (h1dd3n3y3).
First off, find a .yuv video...
This is very important since it is lossless data that we compare. We are encoding, not decoding : )
Then place it in the SourceVideo directory. The scripts will detect it and spew the data in the Data directory... Speaking of scripts...
We have 3 shell executables and 2 python scripts.
The python scripts use plain old matplotlib and numpy
The shell scripts are there to install a virtual environment and automatically run the corresponding script.
For example, run_generate_in_virtual_environment.sh will run the Generate.py script ect.
The run_all_at_once.sh script runs... all at once...
Well actually it just runs the scripts with the right order. First the Generate.py script to generate the logs and conversions of the source video and then the Analysis.py script.
With that in mind, the right way of running these scripts is:
run_generate_in_virtual_environment.sh -> run_analysis_in_virtual_environment.sh or you could run run_all_at_once.sh
If you attempt on running this outside of a virtual environment, please run in this order:
Generate.py -> Analysis.py
You may find sample analysis data in Data/sample_data. It does't contain any video data though in case of any copyright violation...