Sage Kobo (still) can't count?

[Uncle Robin]

One of the main reasons I switched to KOReader as my default was the risibly surreal "stats" offered by Kobo - time per page, time to left to read, none of it ever bore any connexion with reality. Yesterday, I bought a new book and decided to try it in stock. I see nothing much has changed. The sleep cover disagrees with the book's own count my a wide margin, according to the calculator app on my phone. I would have thought those very fundamental figures might agree, at least.

[Sirtel]

Hmm no idea. I'm not even remotely interested in any book stats and never look at them.

[Quoth]

Originally Posted by Sirtel

Hmm no idea. I'm not even remotely interested in any book stats and never look at them.

Agreed. The % left of Chatper and % left of book are OK on Sage and Elipsa, but on all other smaller ereaders a book progess bar is enough.

Easy to see progress by % length of progress bar, like spine position of a paper book.

I do need page numbers on paper books.

[Aramir]

If I had to take a shot. I'd say you read 10% of the actual book. But due to the way the text is reformated / zoomed in / etc depending of your screen and font size the progress bar does not convey actual "page progression" but something more like "screen progression".

Moreover in that logic, "page unit" would have a fix value but "screen unit" would varry depending of the text to display, the font used, the original text format (Carriage Return, empty lines, etc).

tldr : what's on your screen != one page in the book => progress bar doesn't count in pages but in "screen" and depends on many factors making the progression non linear.

[Uncle Robin]

Originally Posted by Aramir

If I had to take a shot. I'd say you read 10% of the actual book. But due to the way the text is reformated / zoomed in / etc depending of your screen and font size the progress bar does not convey actual "page progression" but something more like "screen progression".

Moreover in that logic, "page unit" would have a fix value but "screen unit" would varry depending of the text to display, the font used, the original text format (Carriage Return, empty lines, etc).

tldr : what's on your screen != one page in the book => progress bar doesn't count in pages but in "screen" and depends on many factors making the progression non linear.

I completely understand that the screen unit is not 1 page (Kobo lists the page equivalent as 1005 fwiw), but the total number of screen units is still 100% of the book - if I got to the (in this case) 1593rd screen, there are none left, the book is finished.

Therefore 14% of that number is still 14% of the book, not 10%. An exact match is not to be expected, of course, but the discrepancy persists - in screen count I'm at 21% now, 16% according to the 'sleeping' stat.

Out of interest, I downloaded the epub from Kobo and opened it in ADE. It says I'm on page 191/864 - very near to the percentage count of "screen units' shown in the kepub progress bar, and a figure that makes sense when I've finished 18/55 chapters (excluding end matter, of which there is a lot)


It's a curiosity, not an annoyance, one I'm mildly interested in understanding. Both the Kepub progress bar and the Adobe epub more or less agree on the progress percentage, the Kobo stats don't. It's clear to me that the low stats count is wrong, I'm a lot more than 1/6 of the way through the book, I just find it amusing that Kobo can't even manage to get its own stats counters to agree with each other.

[Aramir]

Therefore 14% of that number is still 14% of the book

Not necessarly, let's say you open a book and at page 13 there is only two lines printed because it's the end of a chapter or whatever. And the page 14 is only haf full
This two lines only page might not be a two lines page on the screen but might be concatenated with the next one. Resulting in one screen holding two IRL pages. Thus using the screen unit you'll progress faster because you're gonna skip all the empty space.

That's an extreme exemple (and maybe even an incorrect one cause chapter might be an exception) but the same might occur with early carriage return and those can add up fast. And I also could be totally wrong, it needs actual investigation to be sure.

[Uncle Robin]

Originally Posted by Aramir

I also could be totally wrong,

Given that that 2 very different page rendering engines (one of which I believe is supposed to mimic paper page count) agree almost exactly on the progress percentage, the above quote seems very likely

[MGlitch]

I'm not sure about epub, but I know for kepub 1 screen does = 1 page. If you want to test this you can increase or decrease any of the text rendering options and see the page count increase/decrease depending on the change.

What I would guess is happening is the sleep cover stats might not be counting some/all of the CSS while the reading progress within the book inadvertently takes the CSS into account because it's so much a part of what makes up the display of text. I'm not sure how much raw CSS is in the book, it'd need to take up roughly 70 screens to get these to align.

Another option is that something specific to the book is wonky, which the kepub tenderer doesn't like.

[JSWolf]

Originally Posted by MGlitch

I'm not sure about epub, but I know for kepub 1 screen does = 1 page. If you want to test this you can increase or decrease any of the text rendering options and see the page count increase/decrease depending on the change.

Sorry, but that is incorrect. KePub does not do 1 screen = 1 page. It used to but no longer. If I am incorrect, then something changed in 4.32.

[MGlitch]

Originally Posted by JSWolf

Sorry, but that is incorrect. KePub does not do 1 screen = 1 page. It used to but no longer. If I am incorrect, then something changed in 4.32.

Please see:

Originally Posted by davidfor

Kepubs use a method that is supposed to be one page per screen. With this, the number of pages changes depending on font size and the other settings. It also means that images are included and the formatting affects the numbering. It seems that the per-chapter numbering is always correct, but, there are problems with the full-book numbering. Most reports are of a page being duplicated. I don't remember reports of them being skipped like you are saying. For this, I am sure that the text formatting and things like images affects the numbering. But, I haven't worked out how. I'm fairly sure that the more complicated it is, the more it is out. And it is probably affected by how long each chapter is. The longer the chapter, the more likely there is a problem. For simple layout in novels, I see a duplicated page number every eight or so pages.

Now you can break this by setting font size and line spacing to their minimums, at which point there is some fluctuation. However going to the other extreme returns to the one page per screen behavior. It should be noted that well before the minimum settings the text becomes unreadable without magnification.

[JSWolf]

Originally Posted by MGlitch

Please see:



Now you can break this by setting font size and line spacing to their minimums, at which point there is some fluctuation. However going to the other extreme returns to the one page per screen behavior. It should be noted that well before the minimum settings the text becomes unreadable without magnification.

Let's just say that it's broken and call it a day.

[davidfor]

It isn't that Kobo can't count, it is that no one knows exactly what they are counting.

Firstly, the method for kepubs and epubs is different.

For epubs, it is probably pretty much just the page count, but, as that is calculated in a questionable way, it isn't perfect.

For kepubs, is not related to the page count at all. During the import, Kobo uses the internal file size to calculate a percentage of the full book for each chapter. That, plus a percentage for the beginning of each chapter, is stored in the database. When reading, the percent read at the beginning of the chapter will be displayed. And as you page through the chapter, the percent read is calculated within the chapter. This, I think, is done using the word count in the chapter. The graphs in the stats popup are done using the word count, so the percentage within the chapter is probably based on that.

And for the record, the word count for each chapter is calculated the first time a kepub is opened. It is stored in the database for use later.

The above is based on experimentation. I discovered some of it when playing with the stored word count. And some of it when doing the ToC updating in the Kobo Utilities plugin. And I have played with the different numbers in the database to see what effect it has. And all the numbers are stored as integers, so there are also rounding errors happening.

[Uncle Robin]

Originally Posted by davidfor

It isn't that Kobo can't count, it is that no one knows exactly what they are counting.

Firstly, the method for kepubs and epubs is different.

I mentioned that fact 2 different renderers offered very similar progress percentages was evidence (imo, ofc) that they were more reliable than the "% figure shown on the sleeping display (and in the book stats under "Activity")


I have been further intrigued to discover an unexpected trend - the gap is getting BIGGER the further I go into the book. I had thought that as I progressed through the book, the 2 divergent stats would move toward each other, but so far, the opposite is true.

When I took those screenshots, the progress bar said 14 % and the "book %" said 10%. Then, in one of my replies above, it was 21% (22% in the ADE epub) versus 16%. Now the progress bar says 27.5% while the "book %" says 21%. If this trend continues, I'll be halfway through my 2nd read before reaching 100% on the sleep cover

[JSWolf]

Originally Posted by Uncle Robin

I mentioned that fact 2 different renderers offered very similar progress percentages was evidence (imo, ofc) that they were more reliable than the "% figure shown on the sleeping display (and in the book stats under "Activity")


I have been further intrigued to discover an unexpected trend - the gap is getting BIGGER the further I go into the book. I had thought that as I progressed through the book, the 2 divergent stats would move toward each other, but so far, the opposite is true.

When I took those screenshots, the progress bar said 14 % and the "book %" said 10%. Then, in one of my replies above, it was 21% (22% in the ADE epub) versus 16%. Now the progress bar says 27.5% while the "book %" says 21%. If this trend continues, I'll be halfway through my 2nd read before reaching 100% on the sleep cover

What % is more accurate?

[Uncle Robin]

Originally Posted by JSWolf

What % is more accurate?

As I've said repeatedly above, it seems safe to assume that if two different page renderers, ADE epub and the Kepub's internal progress bar, report almost identical figures, that number is likely to be accurate. Also, to have completed chapter 27/55 and be at 21%? Unlikely - ADE epub says 285/864 (32%), kepub progress bar says 494/1593 (31%) - either/both more credible