[sdiy] From Bernie of Electronotes

[KA4HJH]

> On Jun 25, 2017, at 3:01 PM, Barry Klein <barryklein at cox.net> wrote:
> 
> It is not easy at all to make EN's into an ebook. OCR sucks, stuff doesn't result in the same pages, and schematics get scrambled. I tried it with my book and it is like writing a new book. My book is 300 pages and I still haven't done it as it just is more work than I want to take on. Now that the custom chips are coming back I'm more motivated though. It is feasible to put these into pdf's, which I have for my books, but the files are large and they don't convert to ebook formats well.

You are correct on all points.

Having done this sort of thing before I can assure everyone that this would be a HUGE task because of the graphics. Text scan with OCR? A cakewalk for a document feeder with duplexing except for some of the odd technical text. The graphics? Well, fortunately Electronotes only has line art and no screened images. That eliminates a lot of problems I won't go into but every page with illustrations would have to be scanned a second time at 600dpi in monochrome with CCIT-whatever compression (the same lossless compression used for FAX). Next, each piece of artwork would have to be eye-balled for problems like illegible spots and rotated if it's seriously out of alignment. Then, each illustration (some pages have more than one) would have to be copied and pasted by hand onto the OCRed page.

Why is all this necessary? Because lo-res jpegs do horrible things to text and subtle details in schematics and block diagrams. Hi-res monochrome is much more legible and actually takes up less space because of the high degree of compression.

If you haven't seen them already take a look at the pdfs in the top two posts on my largely dormant blog:

http://www.astarcloseup.com


These pdfs have OCRed text, and both screened and line art. This required three scans at different settings, cleanup and tweaking of all of the graphics, and placing the finished illustrations one at a time into the pdfs. All done by "hand". Each took hours to complete. All the text had to be proofread and reset in the original typeface with special attention to text like "±15V". If your email app doesn't display that last quote correctly I used the "plus and minus" character, which no OCR program could identify at the time. Electronic articles are full of oddball text like that. Getting "V" followed by a subscript "cc" is something I haven't even tried yet. I don't know if you could "teach" a decent OCR app to recognize that or not. The final text layout would have to be done in something like Quark Xpress or InDesign because of things like subscripts.

Note that the line art in those articles is only 150dpi. We were all on dialup back then so I had to keep the file size down.

The bottom line: Electronotes would be a gargantuan task even if only done half right. The idea of doing the Builder's Guide and the Preferred Circuits Collection alone is doable but would take weeks.


Terry Bowman, KA4HJH
"The Mac Doctor"