Enhancement/Expansion from image analogies or the simple frame-by-frame method of video analogies.
By considering an extra time dimension, t, we are making our patches the shape of a 3D cube.
With this, we want to input 3 videos similar to our image analogies, but just as videos.
Our results would be to get a video that uses the same style as A and A' applied to our B.
Enhancement/Expansion from image analogies or the simple frame-by-frame method of video analogies.
By considering an extra time dimension,
t
, we are making our patches the shape of a 3D cube. With this, we want to input 3 videos similar to our image analogies, but just as videos.Our results would be to get a video that uses the same style as
A
andA'
applied to ourB
.Some references to look at: Video style transfer by consistent adaptive patch sampling.pdf video_analogies.pdf ==> https://nivha.github.io/vgpnn/