# Image to LaTeX

Turn a photo or screenshot of notes, equations, or a table into LaTeX you can edit and compile.

Image to LaTeX is the bridge between something you can see and source you can change. Use it for a photographed whiteboard, a clean equation screenshot, lecture notes, or a table that would be tedious to type again.

It transcribes the image with a compatible Ollama vision model on your device. The image is not sent to a hosted model.

## Use it when

- the wording and structure are visible but not selectable;
- you want editable source, rather than placing the image in a paper;
- the image contains a mixture of prose, math, and a modest table.

Use [Equation to LaTeX](/docs/tools/equation-to-latex/) when the image contains only one expression. Use OCR software designed for bulk documents when you have hundreds of scanned pages.

## Transcribe an image

1. [Turn on Oleafly Tools](/docs/tools/#open-a-tool) if you have not already.
2. Open **Image to LaTeX**, or run `/image-to-latex` from the search bar.
3. Drop a PNG, JPEG, or WebP image into the input area.
4. Choose **Convert**.
5. Read the LaTeX result, then copy it, save `transcription.tex`, or create a LaTeX project.

For example, a photograph containing a heading, a short paragraph, and

```text
∫₀¹ x² dx = 1/3
```

should return editable prose and math rather than a flattened picture. The model is told to preserve the visible wording, equations, tables, and reading order without filling in missing facts.

## What to check

Vision transcription can mistake similar marks: `1`, `l`, and `I`; `0` and `O`; or a minus sign and a dash. Handwriting, glare, low contrast, cropped subscripts, and dense multi-column layouts increase that risk.

Compile the result and compare it with the image. Pay particular attention to:

- limits on integrals and sums;
- superscripts, subscripts, and accents;
- column order in tables;
- labels, units, and punctuation.

:::caution[Input limits]
Images must be PNG, JPEG, or WebP and smaller than 20 MB. You also need Ollama running with a model that supports images. [Set up Ollama](/docs/ollama/) if the tool reports that no vision model is available.
:::

Image to LaTeX transcribes what is present. It does not prove an equation, correct a calculation, or verify that a quoted passage is accurate.