Skip to main content
Intelligence

Understanding AI Photo Analysis

Every photo you upload is automatically analyzed by AI. Here is what it detects, how it powers your experience, and what it means for search and organization.

What happens when you upload a photo

Within seconds of uploading, AI runs a multi-step analysis on every photo. This happens automatically — you do not need to click anything. The analysis powers Photo Insight, search, recommendations, and the restoration flow.

The analysis includes: damage detection (scratches, tears, stains, foxing), face detection (count and positions), blur assessment (motion blur, defocus, camera shake), noise estimation, color analysis (grayscale, sepia, faded), era estimation, capture classification (flatbed scan, phone capture, photo-of-photo), and warp/curl detection.

AI vision enrichment

In addition to computer vision analysis, an AI vision model generates a text description of each photo — who appears to be in it, the setting, estimated decade, mood, and notable objects. These descriptions power the AI-powered search in your library.

The AI also generates tags (up to 10 per photo) covering themes like 'Group Photo', 'Landscape', 'Has Faces', 'B&W', 'Needs Restoration', and more specific descriptors based on what it sees.

AI descriptions are estimates, not facts. The AI does not know who is in your photos — it describes what it sees. You can edit captions and add your own tags at any time.

How analysis powers Photo Insight

Photo Insight combines all analysis results into a condition grade (Good, Fair, Needs Repair, Heavy Damage) with specific damage labels (fading, scratches, tears, stains, glare, blur). Each label points to the tool that can fix that issue.

The analysis also generates a ranked repair plan — safe steps, optional steps, and blocked steps (with reasons) — tailored to your specific photo. For heavily damaged photos, the system may recommend a premium restoration lane with more thorough processing. Conservation guidance is added when your photo may need careful physical handling before AI processing.

How analysis powers search

Every description, tag, and caption generated by AI is searchable. When you search your library for 'wedding', 'beach', or 'grandmother', the results come from AI-generated metadata — not just filenames.

As you add more photos, the search becomes more useful. AI analysis runs on every upload, so your entire library is searchable from day one.

Smart album suggestions

Based on the analysis of your photos — era, mood, themes, tags, and scene types — Nostalgia can suggest albums that group related photos together. These suggestions appear in your library and you can accept, modify, or ignore them.

Smart albums update automatically as you add new photos that match the criteria.

Privacy and data

Analysis metadata (descriptions, tags, face count, era estimates) is stored with your photos and is only visible to you — or to vault members if you share photos to a Family Vault.

Face detection identifies the number and position of faces but does not perform facial recognition (identifying who a person is). Person-to-photo linking is manual and user-initiated.

For full details on how analysis data is used, see the Privacy Policy and AI Policy in the legal section.

See Photo Insight in action

Upload a photo to see the analysis, recommendations, and metadata that power restoration and search.