Note: This page explains what
video-intelligence.jsoncovers, but doesn't list every field, key name, or default. To actually configure VIF, use README.wse-plugin in the VIF GitHub repo as the source of truth for the current field list.
VIF is configured as individual JSON files in the conf.modules directory, one file per stream. Default.json holds the defaults applied to any stream that doesn't have its own configuration file, while each per-stream file overrides those defaults for that specific stream. Together, these cover connection settings for reaching the VIF service, video-processing defaults (frame rate, resolution, grayscale, and similar), object and scene analysis options, the types and delivery modes (immediate, batch, or rollup) available to event listeners, and a set of debugging options for troubleshooting.
A separate vlm_analysis block, configured the same way, specifically controls vision-language model analysis. For more information, see Use VLM analysis in VIF.
A synthetic_analysis block, configured the same way, controls synthetic video detection. For more information, see Use Synthetic Video Detector in VIF.
