HUB
Sign inSign up

braydin72/mediaforge

By braydin72

•Updated 24 days ago

Self-hosted media pipeline: auto-identifies, transcodes, and organizes video into your library

Image
Content management system
0

1.6K

braydin72/mediaforge repository overview

⁠MediaForge

Ingest · Transcode · Organize

MediaForge is a self-hosted media ingest engine. Drop a file in the watch folder and MediaForge handles the rest — it detects the codec, identifies the title via TVDB/TMDB/OMDb, renames it to match your media server's conventions, and routes it straight to your library. H.264 files are automatically queued for hardware-accelerated HEVC encoding first. Nothing lands in your library until it's identified, named correctly, and fully processed.

MediaForge is not a media server — it feeds Plex, Jellyfin, Emby, or any other media server that watches a library folder.

⁠What it does

  • Watches a folder for incoming media files
  • Detects codec via ffprobe (not filename guessing)
  • Identifies movies via TMDB with OMDb fallback; TV shows via TVDB with TMDB and OMDb fallback
  • Verifies matches using an optional LLM (Anthropic, OpenAI, or local Ollama)
  • Renames to Plex/Jellyfin/Emby-compatible naming conventions with configurable templates
  • Routes HEVC files straight to your library; H.264 files to the encode queue first
  • Encodes H.264 → HEVC (or AV1) using hardware acceleration (NVIDIA, AMD, Intel) or CPU fallback, with VMAF-driven quality control
  • Moves finished files to your library, handling cross-device moves cleanly
  • Never fails silently — anything that can't be processed automatically lands in a Review Queue with a specific, human-readable reason

⁠Quick start (docker run)

docker run -d \
  --name=mediaforge \
  -p 8080:8080 \
  -e PUID=99 -e PGID=100 \
  -v /path/to/appdata/mediaforge:/config \
  -v /path/to/incoming:/incoming \
  -v /path/to/cache/staging:/staging \
  -v /path/to/media/Movies:/media/Movies \
  -v "/path/to/media/TV Shows:/media/TV Shows" \
  --restart unless-stopped \
  braydin72/mediaforge:latest

Then open http://[HOST-IP]:8080, add your TMDB/TVDB API keys in Settings, and point the watch folder at your incoming directory.

⁠Quick start (docker compose)

services:
  mediaforge:
    image: braydin72/mediaforge:latest
    container_name: mediaforge
    runtime: nvidia                        # remove if not using NVIDIA
    environment:
      - PUID=99
      - PGID=100
      - NVIDIA_VISIBLE_DEVICES=all         # remove if not using NVIDIA
      - NVIDIA_DRIVER_CAPABILITIES=all     # remove if not using NVIDIA
    ports:
      - 8080:8080
    volumes:
      - /path/to/appdata/mediaforge:/config
      - /path/to/incoming:/incoming
      - /path/to/cache/staging:/staging
      - /path/to/media/Movies:/media/Movies
      - /path/to/media/TV Shows:/media/TV Shows
    restart: unless-stopped

⁠Volumes

PathPurpose
/configApp config, database, and logs (mediaforge.yaml)
/incomingWatch folder — drop new media here
/stagingWorking directory for in-progress transcodes (put on fast storage)
/media/MoviesMovie library output
/media/TV ShowsTV show library output

⁠GPU acceleration (optional)

MediaForge falls back to CPU encoding automatically if no GPU is configured.

  • NVIDIA: run with --runtime=nvidia, set NVIDIA_VISIBLE_DEVICES=all and NVIDIA_DRIVER_CAPABILITIES=all
  • Intel / AMD: pass through --device=/dev/dri and set PGID to the group owning /dev/dri/renderD128 on the host

If hardware encoding fails mid-job, MediaForge automatically falls back to the next available encoder.

⁠Configuration

Configuration lives at /config/mediaforge.yaml (most settings are also editable from the web UI). API keys required:

  • TMDB (required, free) — movie/TV metadata
  • TVDB (required for TV, free) — TV episode metadata
  • OMDb (optional) — last-resort fallback

Tag summary

Content type

Image

Digest

sha256:1aca46605…

Size

260 MB

Last updated

24 days ago

docker pull braydin72/mediaforge