Facet vs PixCull

Facet is a free, local tool for scoring, culling, organizing, and serving a web gallery for photo libraries. It runs on the user's machine without cloud services, accounts, or API keys. Its scoring evaluates nine dimensions, including aesthetic quality, composition, sharpness, color, exposure, face quality, subject saliency, and dynamic range. It detects bursts, similar photos, blinks, and duplicates, with a dry-run preview for automatic culling. The gallery includes natural-language semantic search, filters, timeline and map views, albums, and themed slideshows. Face tools group people and support searching, renaming, merging, and organizing clusters. XMP sidecars provide metadata interoperability with Lightroom, Capture One, digiKam, and darktable. Users can send photos to a WebDAV inbox, or share revocable album links that need no recipient login; proofing supports hearts and comments, with an optional PIN. Core functions work on CPU, while an NVIDIA GPU unlocks additional models and features. Facet can be installed with Docker or natively on Linux and macOS, requires Python, and does not provide RAW editing or development.

PixCull is an open-source photo and video culling tool that sorts shoots into keep, maybe, and cull groups. It scores each frame from 1–5 stars on technical quality, subject, composition, light, moment, and aesthetic, with a written rationale for each area. The product describes nine photography verticals, from weddings and wildlife to journalism and still life. For photo workflows, it exports XMP sidecars for Lightroom and Capture One, IPTC captions, and HTML galleries; it can also watch tether destinations and sync folders through iCloud, Dropbox, or NAS. Video tools score clips, identify reel candidates, transcribe audio, and export EDL files for Premiere or Resolve. RAW decoding, scoring, face analysis, and GPS analysis can run locally. The maker says photos are sent to MiniMax for judging by default since v2.50 after recorded consent; cloud judging can be disabled with `--vlm-mode off`. PixCull states it has no telemetry, and cached models plus disabled cloud judging can allow runs without a network connection. The plan is free; optional cloud model calls may incur API costs.