Video style transfer using Style Transfer by Relaxed Optimal Transport and Self-Similarity (STROTSS)
Each frame of the content video is styled by the given style image. strotss.py produces a directory with each styled frame as well as the original frame. The script requires ffmpeg to stitch the images into a video.
A forked version of https://github.com/futscdav/strotss. See the original code and links to paper at https://github.com/nkolkin13/STROTSS
Usage:
python strotss.py <content> <style> [--weight 1.0] [--output strotss_animation] [--animation_fps 10] [--resize_to 512] [--device "cuda:0"]
