https://github.com/valentinvignal/hdstyletransfer

A HD Style Transfer script
https://github.com/valentinvignal/hdstyletransfer

style-transfer tensorflow tf

Last synced: about 1 month ago
JSON representation

A HD Style Transfer script

• Host: GitHub
• URL: https://github.com/valentinvignal/hdstyletransfer
• Owner: ValentinVignal
• Created: 2020-02-01T17:55:46.000Z (almost 5 years ago)
• Default Branch: master
• Last Pushed: 2022-11-22T06:06:24.000Z (about 2 years ago)
• Last Synced: 2023-03-05T10:53:29.961Z (almost 2 years ago)
• Topics: style-transfer, tensorflow, tf
• Language: Jupyter Notebook
• Homepage:
• Size: 29 MB
• Stars: 2
• Watchers: 1
• Forks: 3
• Open Issues: 4
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# HDStyleTransfer

This repo is a personal project to perform style transfer between

images. The goal was to create HD stylized images.

> The images size can be at least 1024 x 1024 pixels.

As you see in the results I successfully created acceptable stylized HD

images using the GPU of

[Google Colab](https://colab.research.google.com/). :tada:

## Style Transfer

### Results

Here are some results of my project:

| Content Image | Style Image | Result Image |

| :---: | :---: | :---: |

| ![logo_veval_2](doc/images/logo_veval_2.jpg)| ![fractal](doc/images/fractal.jpg) | ![logo_veval_2](doc/images/logo_veval_2_fractal.png)|

| ![tree_road](doc/images/tree_road.jpg) | ![surface_pro](doc/images/surface_pro.jpg) | ![tree_road_surface_pro](doc/images/tree_road_surface_pro.png)|

| ![tree_road](doc/images/tree_road.jpg)| ![fractal](doc/images/fractal.jpg) | ![tree_road_fractal](doc/images/tree_road_fractal.png) |

### Google colab

The file `style_transfer.ipynb` will perform the style transfert

algorithm for a combination of given images and parameters.

The steps for the setup are explained in the file.

### Script

The script `style_transfer.py` will generate one set of stylized images

for one combination of content/style images and parameters

```cmd

python style_transfer.py

```

The content and style images must be in a `content` and `style` folders

or `content.zip` and `style.zip` files at the root of the project

The parameters can be changed by creating and editing the files:

- `style_transfer_options.json`

- `style_transfer_parameters.json`

- `style_transfer_parameters_list.json`

It will create a `results` folder with the stylized images in it.

To apply the style transfer on several images, run the command

```cmd

python style_transfer.py

```

 as many times as you need