b10.studio

Text

Caption Extractor.

Pull the on-screen text out of an image or a video/reel with OCR — captions, subtitles, overlays and stickers. We scan across the whole clip, so text that only flashes for a moment still gets caught.

Guest mode · up to 50 MB per file. Sign in for 500 MB.

30-second walkthrough — pull the on-screen text out of any clip.

Overview

About Caption Extractor (Image & Video OCR)

The free Caption Extractor reads the text baked into your media with OCR — captions, subtitles, hardcoded overlays, stickers and watermarks. Drop in a photo or a video (reel, short, story) and pick a language; for videos we sample frames across the whole clip and merge the recognised lines, so a caption that only flashes on screen for a second is still captured. Everything runs server-side with tesseract — nothing to install.

How it works

  1. 1

    Upload an image or video

    Drop in a JPG, PNG or WebP image, or an MP4, MOV or WebM video/reel that has text on screen.

  2. 2

    Pick the OCR language

    English, French or Spanish — or Multi to read all three at once when a clip mixes languages.

  3. 3

    Copy or download the text

    The recognised captions appear as text you can copy to the clipboard or download as a .txt file.

Frequently asked questions

How do I extract captions from a video?

Upload the video, pick the OCR language, and the tool samples frames across the clip, reads the on-screen text with OCR, and merges the lines so repeated captions only appear once.

Can it read text from an image?

Yes — drop in a photo or screenshot and the OCR engine returns the text it can read, ready to copy or download.

Which languages are supported?

English, French and Spanish, plus a Multi mode that reads all three at once for clips that mix languages.

Is it free and watermark-free?

Yes — OCR is completely free, needs no account, and the extracted text is plain, unbranded text.

Why did it miss some text?

OCR works best on clear, high-contrast text. Stylised fonts, motion blur, low resolution or heavy backgrounds can reduce accuracy — try a sharper source or a different language.