media.ocr

OCR

Reads the text from scanned pages and images and writes it to a column.

Find it under Media › Documents › OCR

Also under Import › Check text

Settings

Document: The PDF or image column to read.
Engine: The OCR engine. Click it to see where it runs and whether it needs setting up.
Language: The language of the documents. Leave it blank if you don't know.
PDF render resolution (DPI): How sharply each PDF page is rendered before reading. 200 by default; raise it for small print.
Keep text position: For vision-model engines: also save where each block of text sits on the page. On by default.
Create searchable PDF: Also save a copy of each PDF with the recognized text in it. Needs an engine that returns text positions.

The recognized text, in the column named in Text result (ocr_text by default).
The position of each block of text on the page, in a second column, when the engine provides it.
A searchable PDF, when Create searchable PDF is on.

The import's Check text step: Runs the same OCR while you import. Use the action for files already in a sheet.
Extract fields: Pulls named facts out of text. OCR only produces the text.

Text that reads well can still have wrong names or numbers. Compare a few pages with the image.
Language and searchable PDF depend on the engine.
A higher DPI takes longer and, for hosted engines, can cost more.