This personal fun project is heavily based on abhiskk/fast-neural-style with some changes and a video generation notebook/script:
- Use the official pre-trained VGG model
- Output intermediate results during training
- Add Total Variation Regularization as described in the paper
The model uses the method described in Perceptual Losses for Real-Time Style Transfer and Super-Resolution along with Instance Normalization. The saved-models for examples shown in the README can be downloaded from here.