StyleGAN2 Webtoon / Anime Style Toonify

Overview

StyleGAN2 Webtoon / Anime Style Toonify

Korea Webtoon or Japanese Anime Character
Stylegan2 base high Quality 1024x1024 / 512x512 Generate and Transfer

Korean language
Stylegan2 를 이용해 고화질 반사실화 및 Webtoon / anime 스타일 캐릭터 생성 및 변환을 시도해 보았고 몇번의 시행착오 끝에 아래와 같이 만들 수 있었습니다.
사람 얼굴 사진 Pretrained Model 에서 webtoon / anime 이미지 4백장정도로 Transfer Learning 과 Network blending 을 이용하였습니다.
기존에 나온 것들은 서양 카툰 디즈니적이거나 반사실화적이지 않거나 동양적인 예쁜 케릭터가 잘 나오지 않아서 사실화같으면서 동양적 예쁘고 멋진 webtoon / anime 캐릭터가 나오는데 목적을 두었습니다.
새로운 동양적이고 예쁜 고화질 캐릭터 생성도 잘 하지만 사람 얼굴 사진을 project 후 나온 latent vectors 를 이용해 이를 새로운 훈련된 모델로 generate 하여 photo face to toon/anime face 로 변환하였습니다.
아직 미숙한 점도 있지만 단순 생성, 변환만이 아니고 latent vectors 를 이용하여 얼굴 움직이기, 웃기, 화내기 등 표정 등도 변경 생성 가능하였습니다.

이후에는 반대로 webtoon / anime 캐릭터 이미지를 사실적 photo로 변경하는 것과 얼굴뿐만 아니라 전체 몸을 생성, webtoon / anime 전체 화면을 생성해 보는 것을 해 보려 합니다. 시간이 되는대로...
더 나아가서는 DALL-E, CLIP 를 이용하거나 이런 방식을 이용해 글을 통해 toon/anime 를 생성하는 것을 진행해보고 싶습니다.
웹툰/애니메이션을 좋아하나 그림 솜씨는 없어도 AI를 이용해서 원하는 캐릭터나 toon/anime 을 만들고 이를 움직이고 말이나 글을 통해서 만들수 있지 않을까. 말로 AI 에게 지시해서 만들어가는 tonn/anime 을 상상해 봅니다.


English Language
I tried creating and converting high-definition reflections and Webtoon/anime style characters using Stylegan2, and after several trials and errors, I was able to create it as follows.
Transfer learning and network blending were used with about 400 webtoon / anime images with the human face photo Pretrained Model.
The existing ones are Western Cartoon Disney, not reflective, or Asian pretty characters aren't well out, so the purpose is to come up with Asian pretty and cool webtoon/anime character.
I am good at creating new Asian and pretty high-quality characters, but I used latent vectors that came out after projecting a human face photo and generated it as a new trained model and converted it to photo face to toon/anime face.
Although there are still some inexperienced things, it was possible to change expressions such as facial movements, laughs, and anger using latent vectors as well as simple creation and transformation.

Afterwards, I will try to change the webtoon / anime character image to a realistic photo, create not only the face but also the entire body, and create the webtoon / anime full screen. As time comes...
Furthermore, I would like to proceed with creating toon/anime through text writing using DALL-E, CLIP, or using this method.
I like webtoon/animation, but I don’t have any drawing skills.
Wouldn't it be possible to create a character or toon/anime you want using AI, move it, and create it through words or writings? I imagine a toon/anime created by instructing AI in words.

Compare to Others

Toonify compare image

Generate toon images

Toonify compare image

Transfer Korea Star Photos to toon Images

Toonify compare image

Edit Face 1(Latent direction move)

Toonify Latent direction

Edit Face 2(Latent direction move)

Toonify Latent direction


Requirements

See StyleGAN2 Requirements

Usage

  1. align image

Copy real face photos in "data_test" directory. It will save face croped and aligned images.

python align_images.py data_test data_img_aligned
  1. project image

It will generate *.npy files and projected_image files in "data_project_gen" directory. Make "pretrained_networks" directory. Download pretrain network "https://drive.google.com/file/d/1-04v78_pI59M0IvhcKxsm3YhK2-plnbj/view?usp=drivesdk" and save in "pretrained_networks" directory.

python project_images.py data_img_aligned data_project_gen --num-steps=1500 --network-pkl="pretrained_networks/generator_star-stylegan2-config-f.pkl" --tmp-dir="./tmp"
  1. toonify

It will generate toonified image files in "data_project_gen" directory and It will add "_toon" to filename. Download pretrain network "https://drive.google.com/file/d/1ENBelwQO-_HWFJk0cBzFvi86wPuYwxSz/view?usp=sharing" and save in "pretrained_networks" directory. Download pretrain network "https://drive.google.com/file/d/1AO42jOsgAdWC1L4lgZzk2-FarniPeilp/view?usp=sharing" and save in "pretrained_networks" directory.

Semi-realistic Toon

python toonify_lnw.py --image_dir=data_project_gen --filename=_toon1.jpg --blendednet="pretrained_networks/wonwizard-bl-anistar-s30r64.pkl"

Drawing Toon

python toonify_lnw.py --image_dir=data_project_gen --filename=_toon2.jpg --blendednet="pretrained_networks/wonwizard-bl-anistar-s120r128.pkl"

Based on nvidia StyleGAN2.
개인적으로만 사용가능하며 허가없이 대외적으로 사용할 수 없음.
These are for personal use only and cannot be used externally without my permission.

You might also like...
Non-Official Pytorch implementation of
Non-Official Pytorch implementation of "Face Identity Disentanglement via Latent Space Mapping" https://arxiv.org/abs/2005.07728 Using StyleGAN2 instead of StyleGAN

Face Identity Disentanglement via Latent Space Mapping - Implement in pytorch with StyleGAN 2 Description Pytorch implementation of the paper Face Ide

Fine-tuning StyleGAN2 for Cartoon Face Generation
Fine-tuning StyleGAN2 for Cartoon Face Generation

Cartoon-StyleGAN 🙃 : Fine-tuning StyleGAN2 for Cartoon Face Generation Abstract Recent studies have shown remarkable success in the unsupervised imag

A colab notebook for training Stylegan2-ada on colab, transfer learning onto your own dataset.

Stylegan2-Ada-Google-Colab-Starter-Notebook A no thrills colab notebook for training Stylegan2-ada on colab. transfer learning onto your own dataset h

StyleGAN2 with adaptive discriminator augmentation (ADA) - Official TensorFlow implementation
StyleGAN2 with adaptive discriminator augmentation (ADA) - Official TensorFlow implementation

StyleGAN2 with adaptive discriminator augmentation (ADA) — Official TensorFlow implementation Training Generative Adversarial Networks with Limited Da

A collection of pre-trained StyleGAN2 models trained on different datasets at different resolution.
A collection of pre-trained StyleGAN2 models trained on different datasets at different resolution.

Awesome Pretrained StyleGAN2 A collection of pre-trained StyleGAN2 models trained on different datasets at different resolution. Note the readme is a

A web porting for NVlabs' StyleGAN2, to facilitate exploring all kinds characteristic of StyleGAN networks
A web porting for NVlabs' StyleGAN2, to facilitate exploring all kinds characteristic of StyleGAN networks

This project is a web porting for NVlabs' StyleGAN2, to facilitate exploring all kinds characteristic of StyleGAN networks. Thanks for NVlabs' excelle

Only a Matter of Style: Age Transformation Using a Style-Based Regression Model
Only a Matter of Style: Age Transformation Using a Style-Based Regression Model

Only a Matter of Style: Age Transformation Using a Style-Based Regression Model The task of age transformation illustrates the change of an individual

Fast Neural Style for Image Style Transform by Pytorch
Fast Neural Style for Image Style Transform by Pytorch

FastNeuralStyle by Pytorch Fast Neural Style for Image Style Transform by Pytorch This is famous Fast Neural Style of Paper Perceptual Losses for Real

For holding anime-related object classification and detection models

Animesion An end-to-end framework for anime-related object classification, detection, segmentation, and other models. Update: 01/22/2020. Due to time-

Comments
  • Model Permission

    Model Permission

    Hi, I've asked for model permission for the two pretrained models, if you could please take a look at that.

    Thanks for the code by the way, everything works great.

    opened by SilviaDeo 0
Owner
AI Research Android & Wizard Dreamer , Middle aged man
null
Face2webtoon - Despite its importance, there are few previous works applying I2I translation to webtoon.

Despite its importance, there are few previous works applying I2I translation to webtoon. I collected dataset from naver webtoon 연애혁명 and tried to transfer human faces to webtoon domain.

이상윤 64 Oct 19, 2022
Transfer style api - An API to use with Tranfer Style App, where you can use two image and transfer the style

Transfer Style API It's an API to use with Tranfer Style App, where you can use

Brian Alejandro 1 Feb 13, 2022
A highly efficient, fast, powerful and light-weight anime downloader and streamer for your favorite anime.

AnimDL - Download & Stream Your Favorite Anime AnimDL is an incredibly powerful tool for downloading and streaming anime. Core features Abuses the dev

KR 759 Jan 8, 2023
Little tool in python to watch anime from the terminal (the better way to watch anime)

ani-cli Script working again :), thanks to the fork by Dink4n for the alternative approach to by pass the captcha on gogoanime A cli to browse and wat

Harshith 4.5k Dec 31, 2022
StyleGAN2-ADA - Official PyTorch implementation

Abstract: Training generative adversarial networks (GAN) using too little data typically leads to discriminator overfitting, causing training to diverge. We propose an adaptive discriminator augmentation mechanism that significantly stabilizes training in limited data regimes.

NVIDIA Research Projects 3.2k Dec 30, 2022
StyleGAN2-ada for practice

This version of the newest PyTorch-based StyleGAN2-ada is intended mostly for fellow artists, who rarely look at scientific metrics, but rather need a working creative tool. Tested on Python 3.7 + PyTorch 1.7.1, requires FFMPEG for sequence-to-video conversions. For more explicit details refer to the original implementations.

vadim epstein 170 Nov 16, 2022
Navigating StyleGAN2 w latent space using CLIP

Navigating StyleGAN2 w latent space using CLIP an attempt to build sth with the official SG2-ADA Pytorch impl kinda inspired by Generating Images from

Mike K. 55 Dec 6, 2022
StyleGAN2 - Official TensorFlow Implementation

StyleGAN2 - Official TensorFlow Implementation

NVIDIA Research Projects 10.1k Dec 28, 2022
StyleGAN2-ADA - Official PyTorch implementation

Need Help? If you’re new to StyleGAN2-ADA and looking to get started, please check out this video series from a course Lia Coleman and I taught in Oct

Derrick Schultz 217 Jan 4, 2023
Pretrained models for Jax/Flax: StyleGAN2, GPT2, VGG, ResNet.

Pretrained models for Jax/Flax: StyleGAN2, GPT2, VGG, ResNet.

Matthias Wright 169 Dec 26, 2022