Guide
Copy text from a phone screenshot
Phone screenshots mix the line you want with status bars, keyboards, notches, and chat chrome. Crop first. Then drop the file into Screenshot2Text and copy what comes back.
Updated September 25, 2026
Why phone screenshots need a different crop
A phone screenshot is rarely just the paragraph you care about. The top holds a status bar with the time, signal glyphs, battery icons, and sometimes a Dynamic Island or notch cutout. The bottom may show a home indicator, a navigation bar, or a full QWERTY keyboard if you captured mid-reply. Chat apps add avatars, reaction piles, floating buttons, and unread badges. OCR does not know which band is the "real" text. It reads sharp edges wherever they appear.
Screenshot2Text runs Tesseract in your browser tab. Drop a PNG or JPG from your phone and recognition stays on your device. We do not upload the file for OCR. Ads stay in the page chrome and never receive your image or the extracted string.
The practical win is a tighter frame. Less chrome means fewer false words mixed into the copy box.
Status bars, notches, and system chrome
Status bar icons look like tiny letters and punctuation to an OCR engine. A battery percentage can become a stray number in your paragraph. Wi-Fi glyphs turn into odd symbols. Carrier names sometimes leak into the first line of output. Notch and island regions leave dark curves that confuse line detection when the crop still includes them.
Before you drop the file, crop away the entire top band. On iOS Photos or Android Gallery editors, drag the handles until only the content area remains. If you already airdropped or messaged the screenshot to a laptop, crop there with any simple image tool. A five-second crop often beats a five-minute cleanup of junk lines afterward.
Screenshot2Text already filters some obvious keyboard and chrome rows, but the filter is a helper, not a guarantee. Giving the engine a clean rectangle is still the strongest move you can make.
Keyboards and input chrome
When the soft keyboard is open, the bottom third of the frame is a grid of letter keys, emoji rows, and suggestion chips. That grid is printed text by design. OCR will happily "read" Q W E R T Y as content. Suggestion chips may echo the words you were typing, which looks almost right until you notice duplicates.
Recapture with the keyboard dismissed when you can. Tap outside the field, hide the keyboard, then take the screenshot again. If you only have the keyboard-on frame, crop hard above the top of the keyboard. Leave a little margin so you do not clip the last line of the message, but do not leave letter keys in the frame.
Autocomplete bars and floating "send" buttons create the same problem at a smaller scale. Crop them out when the body text is all you need.
Chat bubbles, dark themes, and tiny type
Messaging apps wrap text in rounded bubbles on busy backgrounds. Avatars and timestamps sit beside the words. Dark themes put light gray type on near-black panels, which classic OCR finds harder than dark ink on a light page. Screenshot2Text flips many dark regions before recognition, but the source still has to be sharp.
Zoom the conversation in the app before you capture when the type is small. Larger letterforms give the engine clearer edges. If a long thread spans multiple screens, take several screenshots and drop them as a batch, or crop each bubble group separately. One dense block per run is slower but cleaner when the layout is crowded.
Stickers, reaction emoji piles, and link previews add noise. If you only need the message body, crop to that bubble. Leave the sticker for another day.
iMessage and WhatsApp screenshots
iMessage and WhatsApp let you copy text inside the app, but a screenshot of the thread is only pixels. People often screenshot a code, an address, or a one time password when the sender cannot resend it, and retyping invites mistakes.
Dark bubbles need a sharp capture. Night themes can soften light type on a dark bubble. If you cannot read a digit with your own eyes at full zoom, retake the screenshot before you run OCR.
Emoji reactions and sticker overlays can speck the letters next to them. Crop them out when the text you need sits underneath.
Quoted replies and stacked bubbles can come back out of order. Crop one bubble group at a time and treat the result as a draft, not an exact transcript.
Check codes, card digits, and addresses against the screenshot before you paste them anywhere. Bubble fonts make 0 and O, or 1 and l, easy to swap.
PNG vs JPG from the phone
Prefer a PNG when the OS still gives you one. Lossless pixels keep thin UI fonts and punctuation sharper. Many share sheets re-encode to JPG or WebP and soften edges. If your messaging app crushed the file, ask for the original screenshot from the Photos library instead of the compressed chat attachment.
Camera photos of another phone screen are a different problem. Glare, moire, and skew hurt every OCR engine. Use a real screenshot on the device that shows the text whenever you can. The format guide on this site covers PNG, JPG, and WebP tradeoffs in more depth.
A short workflow that works
Open the screenshot on the phone or after transfer to a computer.
Crop out the status bar, notch, home indicator, and keyboard.
Crop away avatars and reaction chrome when you only need the message.
Open the free screenshot to text tool. Pick the language that matches the text.
Drop the cropped PNG or JPG. Wait for local recognition to finish.
Skim the editable box. Fix an obvious misread. Copy or download .txt.
If the text breaks at the end of every line, press Join lines to put the paragraphs back together. Removing line breaks from PDF or OCR text explains when to leave them.
For several related captures, drop them together. Batch mode labels each block with a filename header so you can tell which shot produced which paragraph.
If the screenshot holds a QR code rather than text, scan the QR code from the image instead. OCR reads letters, not QR patterns.
Privacy and what you get back
Recognition runs in your tab. Closing or reloading clears the working copy in memory. We do not keep a server-side history of phone screenshots, because we never held the files. First-time visits may download the OCR engine and language data to your browser cache. Those requests do not include your image.
You get selectable text in an editable box and a plain .txt download. You do not get a searchable PDF with a text layer written back into the file. That limit is documented in its own guide. For most "I need this chat line in an email" jobs, copy and .txt are enough.
If policy says the screenshot cannot leave a managed device at all, confirm your browser offline rules still allow the local reader assets, or use an approved desktop package. Screenshot2Text is built for ordinary phone captures that should stay local.