Convert an image to LaTeX (OCR)

The OCR tool reads an image of math or text, such as a photo of handwritten notes or a screenshot of an equation, and turns it into LaTeX that you can insert into your document. On luzbase.com this is called Photo to LaTeX; each image counts as one photo capture.

Extract LaTeX from an image

1. Open a .tex file in the editor and put the cursor where the LaTeX should go.
2. In the toolbar, click Extract text from image (OCR).
3. In the Extract Text from Image dialog, drop an image onto the box, or click it to choose a file.
4. Click Extract Text. “Extracting text…” shows while the image is read.
5. Check the result under Extracted text.
6. Click Insert. The LaTeX goes in at your cursor.

To try another image, click Remove image (the × next to the file name), then choose a new one.

The result can’t be edited in the dialog. Insert it, then correct it in the editor like any other text.

What you get

• Each displayed formula becomes its own equation environment.
• Inline formulas stay inline, between $…$.
• Text is inserted as plain text.

For example, a photo of the Pythagorean theorem might come back as:

\begin{equation}
a^{2}+b^{2}=c^{2}
\end{equation}

Always read the result before you compile. OCR can misread symbols, especially in handwriting.

Which images work

• Common image formats such as PNG, JPG, WEBP, HEIC, GIF and BMP.
• One image at a time, up to 25 MB.
• Mathematical formulas, and text in English. Text in other scripts, such as Chinese, Arabic or Cyrillic, is not recognised.

PDF files and multi-page documents are not supported. To use a page from a PDF, take a screenshot of it first.

Who can use OCR

The OCR button is shown to people who can edit the project. You need to be signed in.

How captures count

Each image you extract counts as one photo capture. The Free plan includes a limited number of captures per month, and paid plans include unlimited captures. For current numbers, see luzbase.com/pricing.

• The count resets on the 1st of each month (UTC).
• Captures count against the person who runs the OCR, not the project owner.
• A capture is not counted if the file could not be opened as an image, was too large, or if the OCR service failed.
• A capture still counts when the image was read successfully but nothing was found in it, such as a photo with no math.

When you reach your limit, the dialog shows how many captures you have used and when the count resets, with a button to see the plans.

You can run up to 12 extractions per minute.

Error messages

Message	What to do
“Please select an image file (PNG, JPG, WEBP, HEIC, etc.).”	The file is not an image. Choose an image file.
“This file is not an image OCR can read. Use a PNG, JPG, WEBP or HEIC image.”	Save the image in one of these formats and try again.
“This image could not be opened. It may be damaged; try saving it again as PNG or JPG.”	Save the image again and try again.
“Image is too large.”	Use an image under 25 MB, or crop it to the part you need.
“OCR is busy right now. Try again in a few seconds.”	Wait a moment, then click Try Again.
“Text extraction failed. Please try again.”	The OCR service had a problem. Click Try Again. This attempt is not counted.

Related

• Editing tools
• Plans and limits