Files
embycovers/README.md
T
ponzischeme89 e0b51397e7 Initial commit
2026-04-15 09:27:29 +12:00

84 lines
2.4 KiB
Markdown

# Emby Thumbnail Generator
A self-hosted web UI that generates landscape thumbnails from your Emby library posters. Uses AI-powered subject extraction (rembg/U2-Net, runs entirely locally) to isolate characters from poster art, then composites them into widescreen thumbnails with customisable layouts.
## What it does
1. Connects to your Emby server via API
2. Search/browse your movie and TV library
3. Pulls the poster for a selected item
4. Extracts the subject (person/character) using rembg (offline, no external API)
5. Generates a landscape thumbnail with the subject positioned to one side and the title on the other
6. Optionally pushes the generated thumbnail back to Emby as a custom Thumb image
## Templates
- **Subject Left, Text Right** — character on the left, title text on the right
- **Subject Right, Text Left** — character on the right, title text on the left
- **Subject Center, Text Overlay** — character centered with title overlaid
## Background Modes
- **Auto Gradient** — samples dominant colours from the poster and creates a dark gradient
- **Blurred Poster** — darkened, heavily blurred version of the original poster
- **Solid Colour** — pick your own background colour
## Setup
### Docker (recommended)
1. Edit `docker-compose.yml` with your Emby details:
```yaml
environment:
- EMBY_URL=http://192.168.1.x:8096
- EMBY_API_KEY=your-api-key
```
2. Build and run:
```bash
docker compose up -d --build
```
3. Open `http://localhost:8500`
### Manual
1. Install dependencies:
```bash
pip install -r requirements.txt
```
2. Set environment variables:
```bash
export EMBY_URL=http://192.168.1.x:8096
export EMBY_API_KEY=your-api-key
```
3. Run:
```bash
python app.py
```
4. Open `http://localhost:8500`
## Getting your Emby API Key
1. Open Emby Dashboard → Advanced → API Keys
2. Click "New API Key"
3. Give it a name (e.g. "Thumb Generator")
4. Copy the key
## Notes
- First generation will be slower as rembg downloads the U2-Net model (~170MB)
- The model runs entirely offline after first download — no data leaves your network
- Generated thumbnails are cached in the `cache/` directory
- Works with both movies and TV series
- The "Apply to Emby" button sets the generated image as the item's Thumb image type
## Tech Stack
- **Backend:** Python, FastAPI, Pillow, rembg (U2-Net)
- **Frontend:** Vanilla HTML/CSS/JS
- **Deployment:** Docker