Excellent (and free) performance on old Japanese text
I've been using i2OCR to scan Japanese text from a dozen or so early 20th century printed documents for a personal project. The source text is small, somewhat smudged, and written right-to-left using the old (pre-war) kanji, so it isn't the easiest text to do OCR on.
Despite this, i2OCR has been 99% accurate: it can deal easily with vertical, right-to-left Japanese text, and it has been able to read very small, complicated and blurry kanji accurately. For me, at least, I have found it completely free and efficient: I have not had to log in or pay anything to use the service through the website.






