View Single Post
Old 02-17-2025, 07:01 PM   #330
DiapDealer
Grand Sorcerer
DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.DiapDealer ought to be getting tired of karma fortunes by now.
 
DiapDealer's Avatar
 
Posts: 28,705
Karma: 205039118
Join Date: Jan 2010
Device: Nexus 7, Kindle Fire HD
Quote:
Originally Posted by Quoth View Post
No more complicated than regular OCR, as it was essentially OCR and deduplication of the scan elements used by the OCR to create glyph table. The aim being to avoid having to proof at all.
There was a proof stage somewhere between the ocr garbage and the final glyphs that were positioned on the screen. Having taken quite a few of the topaz books apart myself, it's clear that avoiding proofing was not a universal dodge by any means. I had several topaz books that actually rendered (and read) beautifully. Searching for text in the books was miserable, however, since it searched the included, but invisible ocr layer. The final product (glyphs positioned on the screen with some sort of coordinate system) was utterly unusable for searching or conversion to other formats.

The only conversion routines ever produced had to rely upon the underlying, mistake-ridden, ocr as well. So the results were never pretty. There was simply no way to match the glyphs from the matrices to letters in order to reconstruct the text of the book.
DiapDealer is offline   Reply With Quote