Is LrGenius a good alternative to AnyVision for Lightroom keywording?

LrGenius is a strong alternative, especially for photographers concerned about privacy or cost. Both tools add AI-generated keywords to photos inside Lightroom. AnyVision routes images through Google Cloud Vision, which means photos are uploaded to an external server and you pay ongoing API costs. LrGenius can run entirely on your own hardware, using AI models built into the plugin or an existing Ollama or LM Studio server — no cloud upload, no recurring fee. LrGenius also adds semantic free-text search, captions, culling, and face recognition that AnyVision does not provide.

What is the difference between AnyVision and LrGenius for keywording?

AnyVision uses Google Cloud Vision to generate keywords from a fixed taxonomy — the labels are defined by Google's model. LrGeniusAI uses open-vocabulary vision-language models (local or cloud) that can generate descriptive, unrestricted keywords and full captions. You can also tune LrGenius prompts to match your specific workflow or subject matter.

No. AnyVision requires a paid plugin license plus Google Cloud Vision API costs for each image processed. LrGeniusAI is completely free and open-source (AGPL-3.0). Local processing has no per-image cost, whether you use the plugin's built-in AI models or Ollama and LM Studio.

LrGenius vs. AnyVision

Free local-AI keywording vs. Google Vision-powered keyword tagging.

Last updated: 2026-10-19

Quick verdict

Choose AnyVision if you are already using Google Cloud Vision in other workflows and want a straightforward integration that pipes those labels into Lightroom, and cost or privacy are not primary concerns.

Choose LrGeniusAI if you want to keep your photos on your own hardware, avoid ongoing API costs, and get open-vocabulary keywords plus captions, semantic search, culling, face recognition, and taxonomic species identification — all in one free, open-source plugin. LrGeniusAI's local models can match or exceed Google Vision's keyword quality for many photography genres.

Feature comparison

Aspect	LrGeniusAI beta	AnyVision
Primary purpose	Keywording, captions, search, aesthetic scoring, species ID, culling, edit training, AI agents (MCP)	AI keyword tagging via Google Cloud Vision
Lightroom integration	Plugin (runs inside Lightroom)	Plugin (runs inside Lightroom)
AI keywording	Yes — open-vocabulary, prompt-tunable	Yes — Google Vision fixed taxonomy
AI captions / titles / alt-text	Yes	No
Semantic / natural-language search	Yes	Indirect only (via generated tags)
Species identification (taxonomic)	Yes — BioCLIP 2 on-device, with rank & confidence	Indirect — Google Vision labels, no ranks
AI culling	Yes, face-aware (beta)	No
Face recognition	Yes (beta)	No
Learns personal editing style	AI Edit — from your own saved edits (beta)	No
Runs locally / offline	Yes (built in, or Ollama / LM Studio)	No — cloud only (Google Vision)
Privacy	Full local option — photos never leave your machine	Photos sent to Google Cloud Vision
Platform	macOS + Windows	macOS + Windows
Open source	Yes (AGPL-3.0)	No
Aesthetic score (best-photo ranking)	Yes — AestheticSigLIP score (1–9) and genre, set at index time; search, sort and cull by it	Not a documented feature
AI agents (MCP)	Yes — built-in MCP agent bridge for Claude Code, Claude Desktop, LM Studio, Cursor; read-only mode and stop button	Not advertised
Pricing	Free	Paid license + Google API costs

Where each tool excels

AnyVision: direct Google Vision integration

AnyVision passes your photos through Google Cloud Vision to extract labels and write them as Lightroom keywords. For users already integrated into the Google Cloud ecosystem, this is a predictable, API-driven approach with consistent output from a well-known model. The trade-off is ongoing API costs, cloud-only processing, and a fixed vocabulary determined by Google's model rather than your subject matter.

LrGeniusAI: open vocabulary, local or cloud

LrGeniusAI uses modern vision-language models that generate open-vocabulary keywords — descriptive, unrestricted terms tailored to what is actually in each photo. You can tune prompts for your genre (wildlife, wedding, stock) and choose whether to use a local model (free, private) or a cloud API (ChatGPT, Gemini, Claude). Beyond keywording, LrGeniusAI adds semantic free-text search, captions for accessibility, culling, and face recognition — none of which AnyVision provides.

For nature work the difference is sharper still. Google Vision returns labels; LrGeniusAI runs BioCLIP 2 on your own machine and writes an actual taxonomic identification — scientific name, common name, rank, confidence and the full Linnean hierarchy — into searchable Lightroom fields, with no API call and no per-photo cost. Where it is not confident of the species it reports the deepest rank it is sure of instead of guessing a binomial.

See the Analyze & Index guide and the Choosing an AI Model guide for details on selecting and configuring your AI provider.

New in LrGeniusAI: aesthetic scores and AI agents

LrGeniusAI scores every photo for how good it looks (AestheticSigLIP, 1 to 9, plus a genre). You can search with it, list your best photos with no search term at all, and culling uses it to order frames within a group.

LrGeniusAI can also be driven by an AI assistant. Through the Model Context Protocol (MCP), Claude Code, Claude Desktop, LM Studio, Cursor and other MCP apps can search your catalog, cull, rate, keyword, fill collections and change develop settings in Lightroom itself. Everything runs on your computer, and with a local model nothing leaves it. An agent works on its own and can be wrong on many photos at once, so there is a read-only mode and a stop button, and you should back up your catalog first.