Skip to main content

Forge

FileFlows Forge is a companion service designed to offload resource-heavy or specialized media tasks from your standard processing nodes.

You do not usually interact with Forge directly. Instead, specific flow elements automatically hand off processing workloads to Forge behind the scenes when they need advanced capabilities like AI transcription, text extraction, or deep quality analysis.

Deployment & Network Flexibility​

Forge is incredibly flexible and can be deployed wherever it makes the most sense for your home or enterprise network footprint:

  • Colocated: It can run on the exact same machine hosting your primary FileFlows Server or an existing processing Node.
  • Distributed Network: It can be installed entirely on a separate, dedicated machine anywhere else on your local network—as long as your processing nodes have network visibility to reach its IP address.

Why Use Forge​

  • Isolate Resource Hogs: Tasks like AI subtitle generation and VMAF quality calculations will easily peg system hardware at 100%. Moving these to Forge ensures your active processing nodes stay snappy and focused entirely on consistent video transcoding.
  • Maximize Mixed Hardware: This architecture allows you to run lightweight, energy-efficient processing nodes (like Mini-PCs or NAS units) for standard tasks, while routing compute-heavy metrics or AI processing back to a single, high-powered central Forge machine.
  • Seamless Flow Integration: The flow-building experience remains completely unchanged. Your nodes automatically delegate the heavy lifting and receive the standardized output (like filter strings or subtitle tracks) back into the active pipeline.
  • Controlled Environment: Bundles and standardizes specialized media utilities into a single, predictable execution layer without needing complex manual tooling setups on every node.

Available Tasks​

Forge currently handles the specialized background execution for the following tasks:

  • Audio Normalization: Provides multi-pass loudness analysis to generate ideal volume filter configurations.
  • Compute VMAF: Computes perceptual video quality scores to dynamically optimize target encoding pipelines.
  • Generate Subtitles: Uses speech-to-text models to generate fully timed text tracks from raw audio.
  • Image Subtitle Converter: Converts raw bitmap image subtitles into universally readable text .srt files so they can be translated.
  • Subtitle Identification: Analyzes text tracks to accurately tag subtitle languages so you can easily organize or translate your library.
  • Subtitle Translator: Translates text-based subtitle tracks from one language to another using language models.

Installation​

Ready to configure your companion instance?