Paragraphs, columns or lines: layout modes in Read

Recognising the words is only half of text recognition. The other half is putting them back in an order and shape you can use. Read by Zaweba gives you two layout modes for that: Paragraphs and columns, which rebuilds flowing text from the page, and Single column, line by line, which keeps the page's own line breaks.

Choosing the right one saves a lot of tidying afterwards. The same page can come out as neat paragraphs or as a ragged list of short lines, depending on which you pick.

Paragraphs and columns

This mode is for ordinary reading text: letters, articles, reports, book pages. On the page, a paragraph is broken into lines only because the paper is a certain width. You don't want those breaks in your text, so Read joins the lines of a paragraph back together.

It also rejoins words that were split with a hyphen at the end of a line, so "infor-" and "mation" become "information" again. It puts a blank line between paragraphs so they stay distinct.

Columns are handled properly. A newsletter or a journal page with two columns is read one column after another, rather than straight across the page, which would mix the two into gibberish. When you save as .docx, the columns are kept as columns.

Read is also sensible about short lines. A block of short lines, such as an address at the top of a letter, keeps its line breaks, because joining it into one long line would be wrong.

Single column, line by line

Sometimes the line breaks are the point. Poetry, lists of names, code, labels, forms and anything laid out deliberately line by line should come out exactly as it sits on the page. This mode keeps the page's own line breaks, in reading order, and doesn't try to join anything.

It's also a useful fallback when a page has an unusual layout and Paragraphs and columns joins things you'd rather it didn't.

Choosing between them

Start with Paragraphs and columns for most documents; it gives the cleanest result for anything written in sentences. Switch to line by line when the shape of the lines carries meaning, or when you want to compare the text with the page line for line.

Tables are a separate matter. Rows of words aligned across wide gaps (three or more rows) are detected as tables and go into .docx as real tables and into CSV as rows of cells.

How layout carries through to your file

The mode affects every way the text leaves Read. Copied or saved as .txt, you get the joined paragraphs or the kept lines. In a .docx, paragraphs become paragraphs, line breaks stay where they belong, and two-column pages become two-column sections. A searchable PDF keeps the page as it looks, with each word laid where it sits, so its layout matches the picture.

Engines and layout

Layout analysis is one of the reasons Read has Tesseract as well as Windows OCR. In Auto, Read chooses Tesseract for pages with columns, tables, multi-page documents and searchable PDFs, because it analyses layout. The engine used and the reason are shown above the text. If a page came out with its columns mixed up, use Read > Read again with Tesseract. Read by Zaweba keeps all of this on your PC.