Real-ESRGAN-Enhanced-Anime-Diffusion

Generate high resolution and quality anime pictures from texts or existed images.

(Based on Anything V4 and Real-ESRGAN)

Colab demo : Demo

Installation

This project requires: Python >= 3.7 (Recommend to use Anaconda or Miniconda)

clone this repository

  git clone https://github.com/dotmet/Real-ESRGAN-Enhanced-Anime-Diffusion.git

install depencies

  cd Real-ESRGAN-Enhanced-Anime-Diffusion
  pip install -r requirements.txt

Run

  python inference.py

Type python inference.py -h in command line to see more options.

The text is passed by -wd or --words, and this arg should be followed by one sentence or the name of a file which contains text(s). Notice that each text in the file should strictly be 1 line (The text in the file will be automatically splited by the symbol "\n"). Example usages: python inference.py -wd "1girl, beautiful eyes, ", python inference.py -wd prompt_keys.txt

Run Web UI

  python app.py