How to read Japanese text in an image
Short answer
To read Japanese text in an image, first get the text out with OCR, then add readings and meanings. Google Lens copies and translates text. A screenshot scan reads the text and adds furigana, a word table and grammar notes. Pop-up dictionaries and ichi.moe help once you have the text.
Tools for reading Japanese in a picture
This comparison is based on how each tool is designed to work. Features change, so check each tool for its current behavior.
| Tool | Takes an image | Gives text | Reading help for learners |
|---|---|---|---|
| Screenshot scan on this site | Yes | Yes | Furigana, romaji, word table, grammar notes, translation |
| Google Lens | Yes, on phone and desktop | Yes, copy or translate | Translation and text copy. No furigana layer |
| Yomitan or similar browser extension | No. It works on real text on a web page | Only if text exists | Pop-up dictionary on hover |
| ichi.moe | No. Paste text | No | Breaks a sentence into words with readings and meanings |
| Plain OCR websites | Yes | Yes | None, text only |
Step by step: from picture to understanding
- Take a clean screenshot. Capture only the text you want. Crop out artwork. A larger, straighter image reads better.
- Get the text. Use OCR. The Japanese OCR page returns plain lines. The main scan also adds readings.
- Add readings. If you used plain OCR, paste the text into the furigana converter to see how each kanji is read.
- Look up unknown words. Use the sentence breakdown for dictionary forms, or the scan's word table for meanings.
- Verify. Compare the result with the image. Models and OCR engines both make mistakes, and similar kanji are easy to confuse.
Why a picture is harder than text
A pop-up dictionary can only read text that exists on a page. When the words are pixels in a manga page, a game frame or a video, there is nothing to select. OCR has to turn the pixels into characters first. Japanese is harder than most languages for this: thousands of kanji, small kana like ゃ and ッ, long vowel marks and vertical layout.
What each tool is good at
Google Lens is quick for a one-off look-up on your phone. It gives you the text and a translation but is not built around learning each word. A browser extension such as Yomitan is great for web pages and e-books where text is real. ichi.moe is a useful text analyzer when you already have a sentence. Plain OCR sites give text and little else.
This site's screenshot scan sits between these. It takes the picture, then gives the lines with furigana, a per-word table, grammar notes and a translation. It is limited to 3 free scans a day, and the output is produced by an AI model, so it can be wrong.
Tips for manga, games and video
- Manga: scan one bubble or panel at a time. Vertical text is harder than horizontal.
- Games: dialogue boxes are usually clean. Turn off any effects that cover the text.
- Video: pause on a clear frame. Hard-coded subtitles scan better than stylized titles.
- Photos of signs: hold the camera square to the sign, avoid glare.
If you only want the text, see the Japanese OCR page. To learn what furigana are, read what is furigana.
Frequently asked questions
Can Google Lens read Japanese?
Yes, it can recognize Japanese text in photos and screenshots, copy it and translate it. It is aimed at quick look-ups rather than study, so it does not show furigana for each word.
Can a dictionary extension read text in an image?
Not by itself. Extensions such as Yomitan work on selectable text. For images you first need OCR to produce text, or a tool built to overlay recognized text on the picture.
What is the best way to get furigana from a screenshot?
Run OCR and add readings in one step. The screenshot scan on this site does that, or run plain OCR and then paste the text into a furigana converter. Check unusual readings by hand.
Why does OCR confuse similar characters?
Many kanji and kana differ by one stroke, such as シ and ツ or 末 and 未. Low resolution and decorative fonts make it worse. Crop closely and proofread.
Is it safe to upload screenshots?
Only upload what you are comfortable sending to an AI service. On this site images are processed in memory and not stored, but they pass to an AI model provider. See the privacy page for the details.
Last updated: