Quote:
Originally Posted by willus
It would be interesting to see a video of this in action. I did a quick search and could not find anything. Does Duokan actually OCR the text in real time, or does it do something like Adobe's clear type where it vectorizes it? If it OCRs the text, does it try to match the original font? How fast and accurate is the OCR? What happens if the document has scanned equations or parts that don't OCR well?
Both of these "problems" could also be seen as advantages (e.g. k2pdfopt, by running on a PC, does not require the user to jailbreak their e-reader). It depends on what the user wants. The k2pdfopt re-flow engine is implemented on some e-reader software (e.g. KOReader), as markom pointed out.
|
IMO the biggest advantage of running k2pdfopt on PC beforehand is quicker flipping through such reflowed scan, because underpowered e-readers (compared to better tablets) only have to show optimized pdf without much computing on their side.
e.g. I can usually use Sony's Prs-t1 landscape mode for A5 pdfs but if I want to use pen for scribbling I have to leave this zoomed mode for normal view getting back to it after scribbling, also page flipping is not nearly as fast as with using pdf optimized with k2pdfopt's fitwidth mode or reflow mode.
There is Duokan for iOS 7+ too.
Textual pdfs (vectors) are being reflowed without pictures as majority of e-readers (with pdf reflow capability) do, text is searchable, highlightable... .
Scanned pdfs are quickly reflowed.
I have joined three of your two-column templates and this six paged pdf has been reflowed on iPad 4 instantaneously, with pictures included, but some words are joined together as can be seen in attachments.
Here's captured screen of your two-column template(converted beforehand to bitmap without a OCR layer) and couple of increased values for bigger letter size.
Some text has been shown as picture.