What file types are supported?
Common image formats (JPG/PNG/WebP/HEIC and more, including iPhone photos) and PDF files. That said, if you have a PDF, PDF to Text is the better choice — it tries reading the real text layer first (instant, fully accurate) and only falls back to OCR on pages that don't have one. This tool doesn't make that distinction — it runs every PDF page through OCR regardless, which is slower and less accurate than direct extraction. This tool is really for content that isn't a PDF to begin with — phone photos of documents, screenshots, and the like.
How accurate is the recognition?
It depends on how clear the source image is and how complex the layout is — printed text and clear document photos recognize well; handwriting, blurry/tilted/poorly lit photos, or complex tables recognize noticeably worse. This tool is built for quickly pulling out text content, not guaranteed 100% accuracy — proofread the result yourself for anything important.
How long does one image or page take?
Roughly 1–3 seconds per page on a typical laptop, depending on image resolution and how much text it contains. The first use needs to download the recognition engine (a little over ten MB, one time only, cached by your browser afterward) — later processing isn't affected.
Are my images or PDFs uploaded to a server?
No. Recognition runs entirely in your browser (the downloaded recognition engine also runs locally) — file content is never sent to any server, and nothing is kept once you close the page.