https://github.com/awjuliani/Pix2Pix-Film
An implementation of Pix2Pix in Tensorflow for use with frames from films
https://github.com/awjuliani/Pix2Pix-Film
gan pix2pix tensorflow
Last synced: 3 months ago
JSON representation
An implementation of Pix2Pix in Tensorflow for use with frames from films
- Host: GitHub
- URL: https://github.com/awjuliani/Pix2Pix-Film
- Owner: awjuliani
- License: mit
- Created: 2017-01-11T23:59:43.000Z (almost 9 years ago)
- Default Branch: master
- Last Pushed: 2017-03-26T18:10:22.000Z (over 8 years ago)
- Last Synced: 2024-11-17T14:12:05.202Z (11 months ago)
- Topics: gan, pix2pix, tensorflow
- Language: Jupyter Notebook
- Size: 434 KB
- Stars: 166
- Watchers: 19
- Forks: 40
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Pix2Pix-Film
An implementation of [Pix2Pix](https://arxiv.org/abs/1611.07004) in Tensorflow for use with colorizing and increasing the field of view in frames from classic films. For more information, see my [Medium Post](https://medium.com/p/f4d551fa0503) on the project.
Pretrained model available [here](https://drive.google.com/open?id=0B8x0IeJAaBccNFVQMkQ0QW15TjQ). It was trained using Alfred Hitchcock films, so it generalizes best to similar movies.
To generate frames, use `ffmpeg` to resize and create `.png` frames from video.
* To resize video: `ffmpeg -strict -2 -i input.mp4 -vf scale=256:144 output.mp4 -strict -2`
* To generate frames: ` ffmpeg -i output.mp4 -vf fps=10 ./frames/out%6d.png`
In order to 'remaster' frames, run Pix2Pix iPython notebook.
To convert 'remastered' frames back into a video, use:
`ffmpeg -framerate 10 -i frame%01d.png -c:v libx264 -r 30 -pix_fmt yuv420p out.mp4`