Manga Ingest With Upscaling

Overview

MangaIngestWithUpscaling is a Blazor-based web application designed to ingest, process, and automatically upscale manga images.

Features

• Manga ingestion: Ingest Mangas into a library structure useful for Kavita and Komga.
  • Keep track of mangas known under many different titles. This software will not just put them into the same folder, but also change the ComicInfo.xml file to reflect the new title.
  • Change the title of a manga. This may sound simple, but going back and changing the ComicInfo.xml file of a lot of chapters is quite a hassle.
• Image upscaling: Enhance image resolution with MangaJaNai upscaling models.

Usage

1. Set up a library through the UI.
2. Ingest manga by uploading putting the manga into the ingest folder you just configured.
3. Profit!

Installation

The preferred way to run the application is through Docker. Below is an example docker-compose file to get you started with CUDA. See below for ROCm (AMD) support.

version: '3.9'

services:
  mangaingestwithupscaling:
    image: ghcr.io/lolle2000la/manga-ingest-with-upscaling:latest
    restart: unless-stopped
    environment:
      TZ: #your timezone here
      Ingest_Upscaler__SelectedDeviceIndex: 0 # if you have multiple GPUs, you can select which one to use
      Ingest_Upscaler__UseFp16: true # if you want to use fp16 instead of fp32, preferred if you have a GPU that supports it
      Ingest_Upscaler__UseCPU: false # if you want to use the CPU instead of the GPU
      # Kavita integration
      #Ingest_Kavita__BaseUrl: http://kavita:5000 # the base URL of your Kavita instance
      #Ingest_Kavita__ApiKey: #Your API key here
      #Ingest_Kavita__Enabled: True # defaults to false
    volumes:
      - /path/to/store/appdata:/data # for storing the database and logs
      - /path/to/store/models:/models # for storing the upscaling models. 
      # ... other folders you want to be able to access from the container
      - /path/to/ingest:/ingest
      - /path/to/target:/target
    ports:
      - 8080:8080 # the web interface will be available on this port
    #user: '1000:1000' # change the user/group for improved security. Note: The user must be part of the 'video' group and have the correct permissions for its mount points.
    # Make sure you have the nvidia-container-toolkit installed on your host.
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]

If you run a ROCm-compatible AMD GPU, you can use the following docker-compose file:

version: '3.9'

services:
  mangaingestwithupscaling:
    image: ghcr.io/lolle2000la/manga-ingest-with-upscaling:latest-rocm
    restart: unless-stopped
    environment:
      TZ: #your timezone here
      Ingest_Upscaler__SelectedDeviceIndex: 0 # if you have multiple GPUs, you can select which one to use
      Ingest_Upscaler__UseFp16: true # if you want to use fp16 instead of fp32, preferred if you have a GPU that supports it
      Ingest_Upscaler__UseCPU: false # if you want to use the CPU instead of the GPU
      # Rest is same as for the CUDA version
    volumes:
      - /path/to/store/appdata:/data # for storing the database and logs
      - /path/to/store/models:/models # for storing the upscaling models. 
      # ... other folders you want to be able to access from the container
      - /path/to/ingest:/ingest
      - /path/to/target:/target
    ports:
      - 8080:8080 # the web interface will be available on this port
    #user: '1000:1000' # change the user/group for improved security. Note: The user must be part of the 'video' group and have the correct permissions for its mount points.
    # The following lines are necessary to run the container with a ROCm-compatible AMD GPU.
    # See https://rocm.docs.amd.com/projects/install-on-linux/en/latest/how-to/docker.html for more information.
    devices:
      - /dev/kfd
      - /dev/dri
    security_opt:
      - seccomp:unconfined

I do not have an AMD GPU, so I cannot test this. If you have any issues, please open an issue.

Building Prerequisites

• .NET 9.0 SDK or later

Running from Source

1. Clone the repository:

   git clone https://github.com/your-repo/MangaIngestWithUpscaling.git
   cd MangaIngestWithUpscaling

2. Restore dependencies:

   dotnet restore

3. Build the project:

   dotnet build

4. Run the application:

   dotnet run

Configuration

The application relies on appsettings.json for configuration. Modify the connection strings and other parameters as needed. Alternatively, you can use environment variables to override the configuration values.

{
  "ConnectionStrings": {
    "DefaultConnection": "Data Source=data.db;Pooling=false",
    "LoggingConnection": "Data Source=logs.db;Pooling=false"
  },
  "Serilog": {
    "Using": [ "Serilog.Sinks.Console", "Serilog.Sinks.SQLite" ],
    "MinimumLevel": {
      "Default": "Information",
      "Override": {
        "Microsoft.AspNetCore": "Warning"
      }
    },
    "Enrich": [ "FromLogContext" ]
  },
  "AllowedHosts": "*",
  "Upscaler": {
    "UseFp16": true,
    "UseCPU": false,
    "SelectedDeviceIndex": 0
  }
}

Tech Stack

• Frontend: Blazor (MudBlazor components)
• Backend: ASP.NET Core
• Database: SQLite
• Logging: Serilog
• Reactive Programming: ReactiveUI (a tiny bit)

Contributing

1. Fork the repository.
2. Create a new feature branch.
3. Commit changes and push to the branch.
4. Open a pull request.