Where did this come from?

ArizonaGuy · 09-23-2026, 01:01 PM

And how do I get rid of it?

I have an old paper history book that I wanted in eBook form, but that is not available anywhere, so I decided to convert the book myself by scanning the pages of my paper version. The work was a pain in the neck, but I was able to scan all of the pages of each chapter and convert the scans to digital form (as rtf files) online. I then used Calibre to convert the individual rtf files to ePub format.

So far, so good, but Calibre inserted a character at the start of every epub book and I don't know why. It is not a big problem, but it is annoying because I don't know why it appeared and I have had trouble removing it.

I have attached screenshots of the start of one chapter in rtf format and the same in the ePub format. Can anyone tell me why this character showed up and how to prevent it? I can probably go through all of the (many) chapters and do it by editing the output, but if I ever go through the pain of doing this again I would like to avoid the issue.

Thanks.

JSWolf · 09-23-2026, 01:26 PM

What does the scanned page look like?

BetterRed · 09-23-2026, 01:44 PM

It may be too late, but RTF is not a format I would choose to convert from, see ==>> What are the best source formats to convert?

I would have chosen DOCX as a starting format, most (all ?) OCR scanning software can create DOCX, and the calibre DOCX Conversion Input module was written by and is actively maintained by Kovid Goyal. I also know of several other ways to convert DOCX to EPUB.

BR

ArizonaGuy · 09-23-2026, 01:55 PM

Quote:

Originally Posted by JSWolf

What does the scanned page look like?

I have attached a screenshot of the first part of the first chapter of the scanned file. That corresponds to the rtf and ePub screenshots, but that annoying character addition generated by Calibre's conversion of the rtf to ePub appears at the start of every chapter file after the conversion.

I am not sure what the scan page has to do with this as the rtf file is the one that is converted, but I suppose there could be some invisible character embedded.

FWIW - I also converted the rtf file to plain text and there is no such character in the start of the text file.

Click image for larger version

Name: Screenshot 2026-09-23 at 10.50.58 AM.png
Views: 44
Size: 206.1 KB
ID: 226168

JSWolf · 09-23-2026, 01:59 PM

Save the RTF as DOCX and convert that.

ArizonaGuy · 09-23-2026, 02:05 PM

Quote:

Originally Posted by BetterRed

It may be too late, but RTF is not a format I would choose to convert from, see ==>> What are the best source formats to convert?

I would have chosen DOCX as a starting format, most (all ?) OCR scanning software can create DOCX, and the calibre DOCX Conversion Input module was written by and is actively maintained by Kovid Goyal. I also know of several other ways to convert DOCX to EPUB.

BR

The online conversion site that I used only displayed the rtf file. I copied and pasted it into TextEdit on my Mac. I used that site because the scan to text processing was flawless, even when the scanned page was distorted due to how paper book pages don't lie completely flat when processing pages at the start and end of the book. Unless I want to destroy the binding, which I don't want to do.

There is some history here. I downloaded some of Durant's Story of Civilization from archive.org and those files were scanned and had to be converted to text if I wanted to be able to search. I used several conversion websites and every one of them gave me gave me mixed results with some sections of poor translation. Even the big name sites. So when I found this site that produced perfect translations I wanted to use it regardless of how the output was formatted.

JSWolf · 09-23-2026, 02:49 PM

Quote:

Originally Posted by ArizonaGuy

The online conversion site that I used only displayed the rtf file. I copied and pasted it into TextEdit on my Mac. I used that site because the scan to text processing was flawless, even when the scanned page was distorted due to how paper book pages don't lie completely flat when processing pages at the start and end of the book. Unless I want to destroy the binding, which I don't want to do.

There is some history here. I downloaded some of Durant's Story of Civilization from archive.org and those files were scanned and had to be converted to text if I wanted to be able to search. I used several conversion websites and every one of them gave me gave me mixed results with some sections of poor translation. Even the big name sites. So when I found this site that produced perfect translations I wanted to use it regardless of how the output was formatted.

Use calibre to convert from DOCX.

Jaws · 10-01-2026, 01:36 AM

ArizonaGuy, does this also happen with RTFs in which the first element is *not* a graphical element (like the overscore above "Foreword" in your example)? If not, it's probably a glitchy implementation of RTF at the online-conversion-to-RTF end; sometimes an RTF that begins with a graphic can look like it has an extra closure to the graphic's anchor (remember, in RTF graphics are embedded twice in each file), and that would be a hanging "}" at the end of the graphic — which is being interpreted as the first character in the text.

And I grouchily disagree about using the .docx format as the baseline for conversion, at least when there are frequent significant nonlinear elements (like footnotes, multicolumn text, and partial-line-length wrapped-around tables/figure-captions). It's… not stable; for example, it's virtually guaranteed to screw up footnote numbering. Microsoft made some easier-to-implement-for-five-page-business-memo decisions in its custom implementation of XML (that's what a .docx file is) that have had unfortunate consequences for longform works with unpredictable repeating elements. The Durant doesn't have that problem (and it is, in fact, available for purchase in ebook form from Simon & Schuster, at a list price of $165

).

BetterRed · 10-01-2026, 02:09 AM

"Never let perfect be the enemy of “it’ll do” ChatGPT, day before yesterday.

maddz · 10-06-2026, 12:36 AM

As it happens, I have the same issue with scrapes from web pages. Usually I only want the text plus associated formatting, not any advertising guff, so it's simpler to copy the text as is, drop it into TextEdit, remove any links, save as RTF, import into calibre, convert to ePub. I suppose I could get a DOCX output from Pages, but it's a more involved process.

It's short stories mostly, so removing the } isn't much of a chore in the editor.

ArizonaGuy · 10-06-2026, 10:05 AM

Quote:

Originally Posted by Jaws

ArizonaGuy, does this also happen with RTFs in which the first element is *not* a graphical element (like the overscore above "Foreword" in your example)?

Sorry for the delay in responding, but I just returned from a trip and only now had a chance to check.

Yes, it does happen with no graphical element as the first character. I loaded the same rtf file and removed the heading so that the file started with a normal text character and got the same result. Out of curiosity I loaded the rtf file into TextEdit, converted it all to plain text and then did not get the same result. With plain text I got only the actual text in the ePub conversion using Calibre.

ArizonaGuy · 10-06-2026, 10:09 AM

Quote:

Originally Posted by maddz

As it happens, I have the same issue with scrapes from web pages. Usually I only want the text plus associated formatting, not any advertising guff, so it's simpler to copy the text as is, drop it into TextEdit, remove any links, save as RTF, import into calibre, convert to ePub. I suppose I could get a DOCX output from Pages, but it's a more involved process.

It's short stories mostly, so removing the } isn't much of a chore in the editor.

I started using Pages to export an ePub from the rtf and that does seem to have solved the problem. I then run the Page produced ePub through Calibre and get a final ePub.

The one thing that is puzzling me is that I can not seem to get an output ePub that will allow me to change the font size in the ePub reader so perhaps I need to play around with Pages a bit more. I have used a Mac for years, ever since I retired, but never used Pages before so I am a complete novice with it.

JSWolf · 10-06-2026, 10:12 AM

Pages > ePub > ePub is going to give you one awful ePub.

ArizonaGuy · 10-06-2026, 10:19 AM

Quote:

Originally Posted by JSWolf

Pages > ePub > ePub is going to give you one awful ePub.

I haven't actually noticed any problem with the result so far.

Usually I switch between reading using one of my eInk devices and apps and listening using one of the TTS apps like ElevenReader and so far it has worked well.

I am not doubting what you are saying. Only commenting that whatever problem the double processing might cause has not been of sufficient seriousness to cause me any issues at this point.

JSWolf · 10-06-2026, 04:05 PM

Quote:

Originally Posted by ArizonaGuy

I haven't actually noticed any problem with the result so far.

Usually I switch between reading using one of my eInk devices and apps and listening using one of the TTS apps like ElevenReader and so far it has worked well.

I am not doubting what you are saying. Only commenting that whatever problem the double processing might cause has not been of sufficient seriousness to cause me any issues at this point.

Leave out Pages. Use a program like Word or LibreOffice and use styles. Then save as DocX and then convert to ePub with calibre.

09-23-2026, 01:01 PM	#1
ArizonaGuy Connoisseur Posts: 85 Karma: 914 Join Date: Feb 2025 Location: Phoenix metro area Device: Kindle Oasis & Paperwhite, Boox Note, ViWoods AiPaper Mini,Paper2Audio	Where did this come from? And how do I get rid of it? I have an old paper history book that I wanted in eBook form, but that is not available anywhere, so I decided to convert the book myself by scanning the pages of my paper version. The work was a pain in the neck, but I was able to scan all of the pages of each chapter and convert the scans to digital form (as rtf files) online. I then used Calibre to convert the individual rtf files to ePub format. So far, so good, but Calibre inserted a character at the start of every epub book and I don't know why. It is not a big problem, but it is annoying because I don't know why it appeared and I have had trouble removing it. I have attached screenshots of the start of one chapter in rtf format and the same in the ePub format. Can anyone tell me why this character showed up and how to prevent it? I can probably go through all of the (many) chapters and do it by editing the output, but if I ever go through the pain of doing this again I would like to avoid the issue. Thanks. Attached Thumbnails

09-23-2026, 01:26 PM	#2
JSWolf Resident Curmudgeon Posts: 85,326 Karma: 154365101 Join Date: Nov 2006 Location: Roslindale, Massachusetts Device: Kobo Libra 2, Kobo Aura H2O, PRS-650, PRS-T1, nook STR, PW3	What does the scanned page look like?

09-23-2026, 01:44 PM	#3
BetterRed null operator (he/him) Posts: 22,870 Karma: 34252730 Join Date: Mar 2012 Location: Sydney Australia Device: none	It may be too late, but RTF is not a format I would choose to convert from, see ==>> What are the best source formats to convert? I would have chosen DOCX as a starting format, most (all ?) OCR scanning software can create DOCX, and the calibre DOCX Conversion Input module was written by and is actively maintained by Kovid Goyal. I also know of several other ways to convert DOCX to EPUB. BR

09-23-2026, 01:59 PM	#5
JSWolf Resident Curmudgeon Posts: 85,326 Karma: 154365101 Join Date: Nov 2006 Location: Roslindale, Massachusetts Device: Kobo Libra 2, Kobo Aura H2O, PRS-650, PRS-T1, nook STR, PW3	Save the RTF as DOCX and convert that.

10-01-2026, 01:36 AM	#8
Jaws JCL Punch-Card Collector Posts: 126 Karma: 606562 Join Date: Jun 2014 Location: Antarctica Device: Aggressively Device Independent	ArizonaGuy, does this also happen with RTFs in which the first element is not a graphical element (like the overscore above "Foreword" in your example)? If not, it's probably a glitchy implementation of RTF at the online-conversion-to-RTF end; sometimes an RTF that begins with a graphic can look like it has an extra closure to the graphic's anchor (remember, in RTF graphics are embedded twice in each file), and that would be a hanging "}" at the end of the graphic — which is being interpreted as the first character in the text. And I grouchily disagree about using the .docx format as the baseline for conversion, at least when there are frequent significant nonlinear elements (like footnotes, multicolumn text, and partial-line-length wrapped-around tables/figure-captions). It's… not stable; for example, it's virtually guaranteed to screw up footnote numbering. Microsoft made some easier-to-implement-for-five-page-business-memo decisions in its custom implementation of XML (that's what a .docx file is) that have had unfortunate consequences for longform works with unpredictable repeating elements. The Durant doesn't have that problem (and it is, in fact, available for purchase in ebook form from Simon & Schuster, at a list price of $165 ).

10-01-2026, 02:09 AM	#9
BetterRed null operator (he/him) Posts: 22,870 Karma: 34252730 Join Date: Mar 2012 Location: Sydney Australia Device: none	"Never let perfect be the enemy of “it’ll do” ChatGPT, day before yesterday.

10-06-2026, 12:36 AM	#10
maddz Wizard Posts: 1,673 Karma: 30496938 Join Date: Mar 2010 Location: UK Device: Kobo Elipsa 2e, Kobo Forma, iPad, Kobo Touch	As it happens, I have the same issue with scrapes from web pages. Usually I only want the text plus associated formatting, not any advertising guff, so it's simpler to copy the text as is, drop it into TextEdit, remove any links, save as RTF, import into calibre, convert to ePub. I suppose I could get a DOCX output from Pages, but it's a more involved process. It's short stories mostly, so removing the } isn't much of a chore in the editor.

10-06-2026, 10:12 AM	#13
JSWolf Resident Curmudgeon Posts: 85,326 Karma: 154365101 Join Date: Nov 2006 Location: Roslindale, Massachusetts Device: Kobo Libra 2, Kobo Aura H2O, PRS-650, PRS-T1, nook STR, PW3	Pages > ePub > ePub is going to give you one awful ePub.