Closed HandsomeDevilv112 closed 7 years ago
You need to calculate the optical flow, then pass it via the arguments -flow_pattern
and -flowWeight_pattern
. See the setup instructions and stylizeVideo.sh for more details.
That's the one. I completely misunderstood that step. If you have com here after walking the same path as I, try to run makeoptflo, follow the pathway instructions. It wants where your frames are, where it should produce the optical flow data, and two spaces for numbers. Though the way the help text is laid out, it looks like it should be [# [#]], but I got it to work with 1 1 Thank you much. Closing with comment that the problem was located between the keyboard and chair.
Apologies if I've missed something very basic, but I'm a bit out of my experience and I don't know where to go from here. For a brief bit of background, I'm on a 1080, I can make neural style images in about a minute and a half without difficulty, I can do deep dream video with deepdreamanim, and I've been able to run the provided example. (Though I should note that it doesn't reach full iterations on each step) But, despite my best efforts, I come back to "bad argument # 2 to 'add' (sizes do not match)" after the first image is created. I appreciate all the help I can get, and don't be afraid of breakin' it down eli5 style. Thank you