SnapOtter Docker Compose setup on a NAS without a GPU

This German-language tutorial sets up SnapOtter with Docker Compose on a Ugreen NAS without a GPU. The presenter describes it as a self-hosted image editing collection with more than 45 tools and local AI functions, with files processed on the user's hardware. Its focus is editing and batch processing rather than photo backup or gallery management.

The setup creates a persistent Data folder, maps port 1349 and uses restart: unless-stopped. The presenter chooses image tag 1 rather than latest to avoid an automatic move to major version 2. At recording time, he shows 1.17.0 as the version behind latest. First login uses admin/admin and requires a password change.

Examples cover resizing, cropping, format conversion and compression. Reducing one image to 200 KB produces visible artifacts. SVG conversion fails on the teddy image, while a logo conversion succeeds with some detail changes. Pipelines can chain operations for batches, and API keys allow processing without opening the interface.

AI modules need additional downloads: the presenter shows roughly 4 to 5 GB for background removal and 3 to 4 GB for OCR. Image enhancement stalls on his server, and background removal initially drives CPU usage to 100%. After restarting the container, the General preset with fast quality removes the background without a GPU. He also explains resource limits, reduced container privileges and a GPU Compose configuration that requires NVIDIA Container Toolkit.