VKT SnapOCR turns any part of a page into editable text: drag a box over a screenshot, a scanned PDF or a video frame and the recognition runs on your machine with a bundled offline engine — no upload, no account.
Why the browser cannot do it, and the right-click-plus-drag workflow that can. What the box actually captures, how the crop keeps HiDPI detail, and when OCR is the wrong tool.
Read articleWhat "runs offline" has to mean in code: a WebAssembly engine packaged inside the extension, two bundled language packs, one host permission, and the single download that is not your pixels.
Read article20 recognitions a day, English and Simplified Chinese, 10 history entries — and the four things a one-time licence changes. What the free tier keeps intact.
Read articleOCR quality is decided before recognition starts: crop tightly, zoom before capturing, pick the right language, and fix stray errors in the editable result box instead of re-running. Practical habits for the local engine.
Read article