Search text inside screenshots on Mac

Find old screenshots by the words visible inside them. Learn how OCR (optical character recognition) makes screenshot text searchable, what it can find, where it works best, and its practical limitations.

Why screenshots become hard to find

You remember taking a screenshot of an error message, a confirmation code, or a specific Slack conversation. But three weeks later, finding that one image among hundreds is nearly impossible with default tools.

Standard search methods fail because:

  • Filenames are useless. "Screenshot 2026-03-15 at 14.23.01.png" tells you nothing about the contents.
  • Finder does not see inside images. You cannot search for "error code 500" or "invoice #12345" and expect Finder to scan your screenshots.
  • Spotlight does not index screenshot text. macOS Spotlight indexes file names and metadata, not the visual text inside images.

The result: you open screenshot after screenshot, squinting at thumbnails, hoping to recognize the one you need.

What "search text inside screenshots" means

Searching text inside screenshots requires a technology called OCR (optical character recognition). Here is how it works in simple terms:

  1. Text detection. The system scans the image and identifies regions that appear to contain text.
  2. Character recognition. It converts those visual characters into actual text data.
  3. Indexing. The extracted text is stored in a searchable index alongside the screenshot filename and metadata.
  4. Query matching. When you search, the system compares your search terms against the indexed text.

This happens entirely on your Mac. No images are uploaded to cloud services, and no text leaves your device.

How OCR helps with screenshot search

Once OCR text is indexed, you can search screenshots using the actual words visible in them. This changes how you find screenshots:

Search examples that work

Error messages like "Connection failed" or "500 Internal Server Error"
Button labels like "Confirm subscription" or "Upgrade plan"
Invoice or order numbers like "INV-2026-00123"
App interface text like "Settings" or "Export complete"
Chat or email snippets from conversations

Instead of remembering when you took a screenshot, you can search for what was in it.

What screenshot text is easiest (and hardest) to search

OCR technology has improved dramatically, but it still has preferences. Text recognition works best under certain conditions.

Text that OCR finds easily

  • Clear UI text. Standard system fonts, good contrast, straight lines.
  • High-resolution screenshots. Retina captures with sharp text edges.
  • Dark text on light backgrounds. The classic black-on-white combination is ideal.
  • Short labels and headings. Button text, menu items, status messages.
  • Print-like fonts. Sans-serif interface fonts work better than decorative typefaces.

Text that OCR struggles with

Recognized less reliably

Very small text (under 10px in the screenshot)
Blurred or heavily compressed images
Handwritten notes or stylized fonts
Text over complex images or gradients
Screenshots with severe JPEG compression artifacts

Even with these limitations, OCR catches most practical screenshot search scenarios. A screenshot of an error dialog or a settings page is usually recognized well enough to be found later.

How TidyShot supports optional OCR search

TidyShot includes on-device OCR as an optional feature. Here is how it works:

Apple Vision framework

TidyShot uses Apple's built-in Vision framework for OCR recognition. This means the text recognition engine is the same one that powers Live Text and other macOS features—not a third-party cloud service.

Background processing

When you enable OCR, TidyShot processes each new screenshot after it is captured. The recognition happens in the background, so it does not interrupt your workflow. You can continue taking screenshots normally while the app builds the text index.

Combined search

TidyShot's search combines multiple sources:

  • Filename. The renamed file (e.g., "Chrome - 2026-04-01.png")
  • App name. Which application was active when captured
  • Date. When the screenshot was taken
  • OCR text. Words detected inside the image (when enabled)

You can search across all of these with a single query. A search for "chrome error" might find a Chrome screenshot containing the word "error" even if the filename only says "Chrome - 2026-04-01.png".

Optional, not required

OCR is entirely optional. If you prefer not to enable it, TidyShot still provides automatic renaming, clipboard copying, and organized library features. You can turn OCR on or off in Settings at any time.

Practical use cases for OCR screenshot search

Here are real scenarios where searching text inside screenshots proves useful:

Bug reporting and QA

You captured an error message last week and need to include it in a bug report. Search "error" or a fragment of the error text to find the exact screenshot without scrolling through hundreds of files.

Customer support

A customer asks about a conversation you had two months ago. You remember sending them a screenshot of a confirmation. Search for their name or "confirmation" in your screenshot library.

Documentation and tutorials

You are writing internal documentation and need that screenshot of the export dialog. Search "export" or "settings" to surface screenshots containing those interface elements.

Reference codes and numbers

Invoice numbers, order IDs, tracking codes—you screenshot them so you do not lose them. Later, search for the specific number to pull up the associated screenshot.

UI review and design feedback

You captured various states of a design with annotations. Search for "v2" or "feedback" to find the specific iteration you need.

Limitations and privacy notes

OCR is powerful but not magic. Here is an honest look at its boundaries:

Not 100% accurate

OCR occasionally misreads characters, especially with unusual fonts or poor image quality. It might read "l" as "1" or miss punctuation. Search works best when you try partial words rather than exact phrases.

Language support

Apple's Vision framework supports many languages, but results vary. English and other Latin-script languages generally work well. Complex scripts (CJK, Arabic, Devanagari) may have varying accuracy depending on the specific characters and fonts.

Processing time

OCR adds a small processing delay after each screenshot. On modern Macs this is usually under a second, but it is not instantaneous. If you take screenshots rapidly in succession, the queue processes them sequentially.

Privacy assurance

All OCR processing in TidyShot happens on your Mac using Apple's local Vision framework. Screenshot contents are never uploaded to external servers for text recognition. Your sensitive screenshots—passwords, financial data, private messages—stay on your device.

Make your screenshots searchable

TidyShot combines automatic organization with optional OCR search. Find screenshots by filename, app, date, or the text inside them.

Download for macOS

Questions about OCR screenshot search

Does OCR work on screenshots I took before installing TidyShot?

Yes, with a caveat. TidyShot can index existing screenshots in your chosen folder, including OCR processing if enabled. However, this happens when you first set up the app or manually trigger a reindex. It does not automatically process your entire screenshot history in the background indefinitely.

Can I search for partial words?

Yes. TidyShot's search is flexible. Searching "conf" will find screenshots containing "confirmation," "configure," and "conflict." This helps compensate for occasional OCR errors where a character might be misread.

Does OCR use a lot of storage space?

The OCR text index is relatively small compared to the screenshots themselves. A text index for thousands of screenshots typically uses only a few megabytes of storage. You can clear the index at any time from TidyShot's Settings without affecting your screenshot files.

Can I disable OCR after enabling it?

Yes. OCR is completely optional. You can toggle it on or off in Settings. Disabling OCR stops new screenshots from being processed, but previously indexed text remains searchable until you clear the index.