Extract text from an image, formatting kept
Not one flat block of words. The headings come back as headings, the lists as lists, and every block tells you the size, weight, and colour it was set in — then you can put the words back on the picture next door.
New accounts get 10 free credits — no card.
One run costs 1 credit, so that covers your first 10. Signing in is free and takes one click — the tools need an account to run.
Every new run is charged, including the same image uploaded again. Extra steps you ask for — another edit round, a further split — cost extra; a failed run refunds. See credit packs and pricing.
Image to TextMarkdown or TXT
- Upload the image — Ctrl/⌘V works anywhere on the page.
- The page is read into blocks, then each block's size, weight, and colour is measured from your own pixels.
- Copy it, or download Markdown with the headings and lists still in it.
Compare the image with the structured text result

10 blocks · exported Markdown, as renderedImage to text for every kind of page
Anything with readable words. A page with a layout keeps its structure; a photo or a note comes back as plain paragraphs.
- Slides sent as pictures
A slide you only have as an image, where you need the words out of it.
- Screenshots
Text that will not select, where retyping was the only other option.
- Documents
A page you want as Markdown, with its headings still headings.
- Figures with readable labels
Labels inside a chart come back marked as a picture rather than transcribed.
- Photos, signs and handwriting
Read line by line as plain paragraphs, without heading levels or list markers.
- PNG, JPG and WebP
Up to 20 MB. Paste with Ctrl/⌘ V anywhere on the page.
How image-to-text extraction works
- 01Upload the image — Ctrl/⌘V works anywhere on the page.
- 02The page is read into blocks, then each block's size, weight, and colour is measured from your own pixels.
- 03Copy it, or download Markdown with the headings and lists still in it.
Structured text, not one block of words
Plain recognition gives you the words. This gives you the page: its headings, its lists and the type each block was set in.
- Headings as headings
Levels come from the read of the page, not from guessing at font sizes afterwards.
- Lists as lists
List items and paragraph blocks, in reading order.
- Size, weight and colour per block
Measured from your own pixels at the image's resolution.
- Markdown
The structure kept: heading levels, lists and paragraphs.
- TXT
The words in the same order, without the structure.
- Words back on the picture
Put the extracted words back on the image next door with the text editor.
Retype, recognise or extract: three ways to get text out of an image
Retyping is exact but slow. Plain recognition is fast but flat. Extraction keeps the structure and the type.
| What you need | Extract with EditableImage | Plain OCR | Retype by hand |
|---|---|---|---|
| Structure | Headings, lists and paragraphs | One block of words | As you type it |
| Type details | Size, weight and colour per block | None | None |
| Figure labels | Marked as a picture | Read as loose words | Skipped or typed |
| Time per page | 20–45 seconds, then a proofread | Seconds | Usually far longer |
A report image with ten text blocks through the extractor, block by block: Image to text with formatting: a 10-block OCR test
Image to text at a glance
Image to Text by EditableImage: Extract text from an image while keeping headings, lists, paragraphs, and measured type details such as size, weight, and colour. Export Markdown or TXT.
- Input
- PNG, JPG, or WebP, up to 20 MB
- Output
- Markdown or TXT
- Best with
- Slides, screenshots, documents, and figures with readable labels
- Processing time
- Usually 20–45 seconds
- Cost
- 1 credit per run; new accounts get 10 free credits, no card
- Your upload
- Never published; the server copy is deleted after 24 hours
Last updated 6 November 2026
When image-to-text extraction helps
- A slide or a document you were sent as a picture, where you need the words out of it.
- A screenshot whose text will not select, and retyping it is the only other option.
- A page you want as Markdown, with its headings still headings.
Image to text questions
Yes. The heading levels and the lists come from the read of the page rather than from guessing at font sizes afterwards, so a subheading comes back as a subheading even where it is set smaller than the body around it.
Yes. The words come back as text you can select, copy, and change — block by block on the page, or downloaded as Markdown or TXT — instead of a picture of words. To keep the text where it sits on the image, as editable text boxes over the original design, use Image to PPT instead.
Yes. Paste or upload the original screenshot at full resolution, then check the block order and small labels before copying or downloading Markdown or TXT.
Yes, for every block, measured from your own pixels at the image's resolution — the size, weight, and colour it was set in, reported next to the words.
Labels drawn inside a diagram or a chart come back marked as a picture rather than transcribed — that is what Image to Draw.io is for. And proofread your numbers: a zero in a display face is read as the letter O often enough to be worth a glance.
Yes. Where the image is not a page with a layout, such as a photo of a sign, a whiteboard, or handwriting, its words are read line by line and come back as plain paragraphs, without heading levels or list markers.
Markdown keeps the structure the read found: heading levels, list items, and paragraph blocks in reading order. TXT keeps the words in the same order without the structure. Neither keeps the page as positioned boxes, and figure labels come back marked as figures.
No. Each block reports size, weight, and colour measured from your pixels — usually enough to restyle the words — but the typeface name and its font file are not recovered.
Start free — signing in gives you 10 credits
No card required. Enough for 10 text extractions, before you decide whether any of this is worth paying for.
- Signing in takes one click and costs nothing. The tools need an account to run.
Image to text and OCR guides
- Image to text with formatting: a 10-block OCR test
We tested a report image with headings, bullets, a paragraph, chart and footer. Nine text blocks kept their structure; the chart stayed a figure.