The following is a heavily modified version of the original project which focuses on the speed
of runtime. Some staggering code clutter was also removed (a lot still remains but should be easier to read now). Additional changes could be made if additional
speed is required specifically the use of pandas
as well as multiprocessing
pool focusing specifically on the DNA evolution and
sampling from DNA will result in the greatest additional boost in performance as numpy.choice
takes 1/3 of the remaining runtime.
Examples of generated images:
It also supports user-created sampling masks, in case you'd like to specify regions where more brushstrokes are needed (for example: to allocate finer details)
- Run the main script
python3 main.py
The following fork makes use of conda to avoid dirtying the environment
- Building conda environment
conda env create --name genetic-drawing --file environment.yml
- Starting env
conda activate genetic-drawing