Peakto introduces powerful AI-driven tools to deduplicate and cull images across your entire library. Unlike traditional tools limited to a single folder or catalog, Peakto lets you run culling and deduplication globally, across collections, sources, external drives, and catalogs.
These tools help you quickly identify duplicate, near-duplicate, or similar images—and decide which ones to keep, flag, or remove.
Select the scope you want to work on:
An entire library
One or multiple collections
A single source or a group of sources
Click the Deduplication & Culling icon (loop icon) in the left panel.
Peakto displays the number of images that will be analyzed before launching the process.
Peakto provides templates that define how images are grouped and evaluated.
Groups nearly identical images
Ideal for removing exact or very close duplicates
Groups similar images (bursts, variations, reshoots)
Designed to help you select the best shots within a series
Use the Similarity slider to adjust grouping behavior:
Move left → stricter similarity (deduplication)
Move right → looser similarity (culling)
Each template is driven by rules that determine which image should be picked or rejected within a group.
Common rules include:
File type (e.g. prioritize RAW over JPEG)
Image size / resolution
Rating
Face quality (for portraits)
Aesthetic score (for general culling)
You can:
Enable or disable rules
Change the order of priority
Apply rules automatically or manually
Different templates emphasize different criteria:
Deduplication typically prioritizes file type, size, and rating
Portrait culling prioritizes face quality
Collection culling prioritizes aesthetic score
After computation, Peakto creates clusters:
Each cluster contains images considered similar based on the selected similarity level
Images without duplicates or similar counterparts are excluded from this view
Clusters are the foundation of the culling and deduplication workflow.
Within each cluster, you can:
Pick one or multiple images
Reject unwanted images
Each cluster displays:
A local summary of picks and rejects
A global summary at the top showing totals across all clusters
Selections can be made:
Manually, image by image
Automatically, by applying rules to all clusters
To review images more closely:
Select a cluster
Click Compare
View images side by side (A/B comparison)
Zoom in to inspect details (depending on image availability)
If original files are online, higher-resolution zoom is available.
When running portrait culling:
Images are grouped by similarity
Face quality becomes the primary rule
You can refine similarity and re-review clusters at any time