Movie Trailers to Comic Strips of Specific Artistic Styles
Functions of the scripts:
pyscenedetect_command.md: use PySceneDetect to extract content-aware frames;
jpeg_res.py: pick the top 6153 photos from flickr30k dataset sorted by abs(width/height-1920/1080) value;
resize_frames.py: use FFmpeg to resize all input images to 256x256
train & test CartoonGAN or use a pretrained model to generate frames of specific artistic styles
img2comicstrips.py: lay out the result frames like in comic strips, insert subtitles at the right place and also page numbers.
Some results:
