How to use it
Drop a picture on the page, paste one with ⌘V or Ctrl+V, or press Choose a file.
The first time, the page gets its text reader, which can take a few seconds. Later pictures are quick.
The words appear in the box next to your picture, and each line it found is lightly boxed on the picture.
Check them against the picture. You can fix a word right in the box.
Press Copy text, or Download .txt to keep them as a file.
What it reads well, and what it doesn't
It reads clear text well: a screenshot of a web page, an error message, a document on screen. A dark-mode screenshot is fine too: the page turns it round before reading, since light text on dark reads worse. It does less well on a photo of a page that's curved, shadowed or at an angle, on tiny text, on handwriting and on decorative lettering. A table comes out as lines of text, not as a grid.
If a picture is hard to read, the page says so under the box. Always read the result before you rely on it, especially numbers and names.
Copy text as you capture
VizShot does this on your own Mac or PC, with the system's own text recognition: press ⌘⇧0 on a Mac or Ctrl+Shift+0 on Windows, drag over any text on your screen, and the words are on your clipboard. It reads the languages your computer reads, a table keeps its columns, and nothing is uploaded. Copy text from a screenshot has the steps.
Questions
Is my picture uploaded?
No. The picture is opened and read inside your browser, and no request carries it anywhere. The page fetches its text reader from vizshot.app, once, and that's all it fetches. To see for yourself, open your browser's network panel and do the whole job.
Why does the first picture take longer?
The text reader is about 5 MB, so the page fetches it only when you drop a picture in, not when the page opens. Your browser keeps it, so the next picture is quick.
Which languages does it read?
English. Text in another language may come out wrong, and the page can't tell you when it has.
Why are some words wrong?
Text recognition guesses from the shapes it sees. A sharp, large, dark-on-light picture gives it the most to go on. A blurry photo, tiny text or an unusual font gives it the least. That's why the box is editable and the picture stays next to it.
Can it read handwriting?
Not well. It's built for printed text.
Can it keep a table or columns?
No. The text comes out line by line, so a table loses its grid and columns can run together. For a table, read one column at a time: crop the picture to that column first.
Do I need an account?
No.
I've just taken a screenshot. Is there a quicker way?
Yes, if you have an app. Copy text from a screenshot shows how VizShot does it in one shortcut, with the words on your clipboard before you've switched windows.