MobileRead Forums - View Single Post - .azw1 or .tpz, antiquated Topaz sample file

Quoth · 10-13-2025, 07:40 AM

Topaz is a special kind of OCR where instead of trying to convert to text, each glyph is stored and similar ones replaced by one in a table.
Unlike a scan encapuslated as a PDF, the table of glyphs is used to build the reflowable page. The advantage over full OCR is that if the matching is set so there are almost no false positives (which would be abysmal for regular OCR) there is no need for human proofing.

It will result in a huge table compared to proofed full OCR and a subsetted Unicode characterset, but will work with no training and any font/alphabet.

Quality is poor compared to human proofed OCR or a decent scan, but worked on the 167 dpi 6" viziplex 4 level screen on K1.

So the only viable conversion is to an image /TIFF/PDF image (calibre won't do it). Then the image might be OCRed.

The K 3 probably can read topaz.

I've no idea how the DRM worked. It may have used mobipocket or more likely Amazon mobi drm.

This (with loads of nasty 3rd party javascript) lists topaz titles. I suspect amazon has removed most now. https://www.kboards.com/threads/the-...az-format.171/

I doubt Calibre scramble plugin works with Topaz. Topaz has no text.

EDIT
see https://wiki.mobileread.com/wiki/Topaz

10-13-2025, 07:40 AM	#2
Quoth Still reading Posts: 14,898 Karma: 110507267 Join Date: Jun 2017 Location: Ireland Device: All 4 Kinds: epub eink, Kindle, android eink, NxtPaper	Topaz is a special kind of OCR where instead of trying to convert to text, each glyph is stored and similar ones replaced by one in a table. Unlike a scan encapuslated as a PDF, the table of glyphs is used to build the reflowable page. The advantage over full OCR is that if the matching is set so there are almost no false positives (which would be abysmal for regular OCR) there is no need for human proofing. It will result in a huge table compared to proofed full OCR and a subsetted Unicode characterset, but will work with no training and any font/alphabet. Quality is poor compared to human proofed OCR or a decent scan, but worked on the 167 dpi 6" viziplex 4 level screen on K1. So the only viable conversion is to an image /TIFF/PDF image (calibre won't do it). Then the image might be OCRed. The K 3 probably can read topaz. I've no idea how the DRM worked. It may have used mobipocket or more likely Amazon mobi drm. This (with loads of nasty 3rd party javascript) lists topaz titles. I suspect amazon has removed most now. https://www.kboards.com/threads/the-...az-format.171/ I doubt Calibre scramble plugin works with Topaz. Topaz has no text. EDIT see https://wiki.mobileread.com/wiki/Topaz