How to find visually similar shots in a video

How visual similarity search works

The Find Similar Looks scan turns your reference frame into a compact visual fingerprint that captures its content, color, and composition, then compares it against fingerprints generated from sampled frames across your video. Frames whose fingerprints are close enough to the reference are grouped into segments.

This is fundamentally different from object detection or facial recognition: it doesn't need to know what is in the frame. It finds frames that simply look like the reference, whether that means similar lighting, color palette, framing, or subject matter.

Step-by-step

1. Open your video with ⌘O.
2. Pause on the reference frame. Navigate to any frame that represents the look you want to find more of.
3. Open Find Similar Looks from the toolbar or Edit → Find Similar Looks.
4. Set your reference. Click Use Current Frame, or load an external image from disk with Choose Image.
5. Adjust the similarity threshold. A lower threshold casts a wider net; a higher one requires closer visual matches.
6. Scan. Segments appear on the timeline as matches are found.
7. Review and export with ⌘E.

What it's good for

• Collecting all shots with a specific lighting setup (golden hour, overcast, indoor warm) across a long shoot
• Finding every wide establishing shot in a documentary
• Pulling close-up cutaways that match a specific frame composition
• Identifying recurring visual motifs in long-form footage
• Matching B-roll shots to a hero frame for visual consistency