Can one popup dictionary work on both web pages and video hardsubs?
YuzuLingo · Nov 1, 2026
Yes. A popup dictionary can cover both surfaces when page text is scanned under a modifier-key hover and burned-in video lines are turned into real overlay text with OCR. The same installed dictionaries, deinflection, speaker, and export path then serve articles and hardsubs.
What classic page scanners already do well
Tools in the Yomitan family read characters that already exist in the DOM. Hold a modifier, move the pointer, and the longest dictionary match opens with readings and glosses. That flow works on blogs, news, and manga viewers that render real text.
Where page-only scanners stop
Hardcoded Netflix, YouTube, Bilibili, and anime lines are painted into frames. There is no text node under the pointer. Installing more dictionary zips does not create characters from pixels, so a page-only scanner never sees the subtitle.
OCR overlay plus the same popup
An OCR overlay crops the subtitle region, recognizes the line on device, and draws the string as selectable text on top of the player. Hover on that overlay uses the same popup you use on the page. Soft subtitle tracks and transcripts can feed the same path when the site already exposes text.
Shared language packs and deinflection
Learning-language packs decide which scripts to try when a line could be Japanese, Korean, Chinese, or another installed language. Client-side deinflection maps conjugated forms back to dictionary headwords so verbs and adjectives still hit the glossary without a server round trip.
Coexistence with other hover tools
Keeping OCR output as real DOM text means Yomitan, 10ten, Zhongwen, and similar extensions can still scan the overlay if you prefer them. A built-in page scanner is optional for readers who want one extension for both articles and video.
