PDFRead 1.8.2 released!

[CheriePie]

So, the prc file generated with the iLiad-p profile should be able to be loaded into the Kindle 2, right?

I'm expecting my Kindle 2 to arrive next week and in preparation I'm converting a few of the PDFs I want to move over. When I used the prc-mobi-p profile, besides the expected downgrade in graphics quality (b/w vs 16 shades gray), the resultant output file had large margins on the bottom and right-hand side. With the iLiad-p profile, each page image came closer to filling the entire page and looked much better.

So I should have no problem pulling this iLiad version into the Kindle 2 right?

[nrapallo]

Originally Posted by CheriePie

So, the prc file generated with the iLiad-p profile should be able to be loaded into the Kindle 2, right?

I'm expecting my Kindle 2 to arrive next week and in preparation I'm converting a few of the PDFs I want to move over. When I used the prc-mobi-p profile, besides the expected downgrade in graphics quality (b/w vs 16 shades gray), the resultant output file had large margins on the bottom and right-hand side. With the iLiad-p profile, each page image came closer to filling the entire page and looked much better.

So I should have no problem pulling this iLiad version into the Kindle 2 right?

By default, they both create a .prc "ebook" but with differing pages sizes and shades of gray i.e.

       prc-mobi  :   520   640   landscape      right           4     gray        prc 
       prc-mobi-p:  520   640   portrait        none            4     gray        prc 
       iLiad        :   768   935   landscape      right          16     gray        prc 
       iLiad-p     :   768   935   portrait        none           16     gray        prc

You can override the above defaults for either profile to suit your needs. I think the iLiad .prc may be a bit too large for the Kindle 2 screen. I'd rather you not have to shrink them down to accomodate, as this adds another (post-) image conversion step that may diminish the quality. It would be better just to increase the default colors to 16 from 4 and use the mobi-prc profiles.

When you get your K2, just experiment which works better (and, of course, report your findings back here )

[CheriePie]

Originally Posted by nrapallo

When you get your K2, just experiment which works better (and, of course, report your findings back here )

Will do. It should be here Wednesday or Thursday. Thanks Nick!

[CheriePie]

Luving ♥ my Kindle 2. Though I still haven't been able to get the particular PDF eBook I've been trying to convert into a readable format that I'm comfortable with.

I couldn't get into the landscape way of reading it... my hands all weren't in the right place for paging and all, and sometimes it split the page mid line, so that actual words were cut in half, so obviously not an option I wanted to continue to pursue.

So working in portrait mode, I tried various settings of widening the vertical size to 600 which didn't seem to make much difference. The only thing that did help somewhat is changing the cropping value from 2 to 1. Because if I can get rid of the margins all along the sides of this thing, then it might be good. But I couldn't go less than 1, so it's still got a good 1" margin all around the text, making the overall text size too small to be comfortably read.

So any suggestions on how to get it so that all the white space on the sides is kaput? Since the screen resolution of the Kindle 2 is 800 x 600 (wide), should I be plugging in these parameters for the size. (I actually didn't notice much difference from when I did that versus leaving them at their default for the prc-mobi-p profile.

[nrapallo]

Originally Posted by CheriePie

Luving ♥ my Kindle 2. Though I still haven't been able to get the particular PDF eBook I've been trying to convert into a readable format that I'm comfortable with.

I couldn't get into the landscape way of reading it... my hands all weren't in the right place for paging and all, and sometimes it split the page mid line, so that actual words were cut in half, so obviously not an option I wanted to continue to pursue.

So working in portrait mode, I tried various settings of widening the vertical size to 600 which didn't seem to make much difference. The only thing that did help somewhat is changing the cropping value from 2 to 1. Because if I can get rid of the margins all along the sides of this thing, then it might be good. But I couldn't go less than 1, so it's still got a good 1" margin all around the text, making the overall text size too small to be comfortably read.

So any suggestions on how to get it so that all the white space on the sides is kaput? Since the screen resolution of the Kindle 2 is 800 x 600 (wide), should I be plugging in these parameters for the size. (I actually didn't notice much difference from when I did that versus leaving them at their default for the prc-mobi-p profile.

It's best to use landscape mode for conversions of PDF to small screen ebook readers. Make sure you try using the 'mobi-prc' Profile, as the other profile 'mobi-prc-p' is for portrait mode. I usually set the Layout mode to 'landscape-half' which ensures at most two page flips for each pdf page. You may want to try 'landscape-third' if there remains too much white space/margins.

The problem with cropping is that it stops at the first major black text/page number/spot on the pdf page. Then the program takes that cropped image and splits it where the Layout mode indicates. The resulting images may/will have text cut in the middle, but there is a bit of overlapping to each image so you don't really loose any lines of text.

Experiement on a few pages only, like Pages 7 to 8, exactly like I showed in this post. Enter the same options and then when it works for a few pages, convert the entire pdf ebook.

If you want me to try to convert it, post it here, if you can, or email me at nrapallo [at] yahoo.ca .

[AnthonyPaulO]

Great job with this app! I just received my Kindle 2 yesterday and have been converting PDF eBooks using your program and it works great! When viewing on the Kindle 2, I am able to read the books albeit the text is really really small. Great job though! Now, as the previous poster was saying, would it be possible to throw in a margin-cropping option so that someone can look at their book and determine that they would like to chop off, say, an inch margin from left and right, and 1.5 inches from top and bottom? I'm talking about just chopping it off regardless of whether there's a page number or footer or whatever, that way it will stretch the image up to Kindle 2 resolution and appear bigger, and if specifying inches is too hard then pixels would work too.

Cheers!

Anthony

[nrapallo]

Originally Posted by AnthonyPaulO

Great job with this app! I just received my Kindle 2 yesterday and have been converting PDF eBooks using your program and it works great! When viewing on the Kindle 2, I am able to read the books albeit the text is really really small. Great job though!

Thanks, but the real praise should go to the original developer of the program, ashkulz! I'm just the maintainer (and it's number one fan!)...

Now, as the previous poster was saying, would it be possible to throw in a margin-cropping option so that someone can look at their book and determine that they would like to chop off, say, an inch margin from left and right, and 1.5 inches from top and bottom? I'm talking about just chopping it off regardless of whether there's a page number or footer or whatever, that way it will stretch the image up to Kindle 2 resolution and appear bigger, and if specifying inches is too hard then pixels would work too.

Cheers!

Anthony

Margin cropping and general cleaning-up of .pdf scans are already supported via the unpaper option box (just insert -v to activate it and supply other paramters there as well). Click that blue "unpaper" text on the PDFRead GUI screen and it will take you to the unpaper website. I haven't worked through all the options, but do know it's possible (the margins are in absolute pixels, I think). Be sure to also read the unpaper - User Documentation.

Perhaps we can try experimenting on a sample .pdf book scan (any suggestions?) to ensure all can help and contribute with this exercise. It's a very time consuming option though, so limited page ranges are recommended while testing.

[AnthonyPaulO]

Hi!

Was just checking out the documentation and I believe that the way to go would be to specify an absolute mask using --mask and specifying the sizes and then disabling the border detection using --no-border-scan since the mask already strips out the border. This would only work for PDF's whose pages are all uniform with respect to their margins. For those cases where pages vary in margins then another approach is needed where the mask remains auto-detected but the border detection is tweaked so that it would ignore page numbers and such. In any case I'm on my iPhone so I can't put anything to the test right now but when I get home I'll run some test and post my findings... I have to first check out the proper syntax for specifying the sizes for --mask.

Regards,

Anthony

Anthony

[nrapallo]

Originally Posted by AnthonyPaulO

Hi!

Was just checking out the documentation and I believe that the way to go would be to specify an absolute mask using --mask and specifying the sizes and then disabling the border detection using --no-border-scan since the mask already strips out the border. This would only work for PDF's whose pages are all uniform with respect to their margins. For those cases where pages vary in margins then another approach is needed where the mask remains auto-detected but the border detection is tweaked so that it would ignore page numbers and such. In any case I'm on my iPhone so I can't put anything to the test right now but when I get home I'll run some test and post my findings... I have to first check out the proper syntax for specifying the sizes for --mask.

Regards,

Anthony

Anthony

I did try experiementing with unpaper when I was updating PDFRead to v1.8, but I can't remember which parameters worked better and what absolute pixels sizes I used for mask/border/threshold. I do remember getting good results from it, though, so I do think trying to figure this out will be worth it!

I'll help you, if I can. Good luck!

[AnthonyPaulO]

Hi!

You mentioned having to type in -v for unpaper and I was wondering whether you were only referring to the command-line usage or if I have to start my unpaper switches in the GUI's textbox with -v.

Regards,

Anthony

[nrapallo]

Originally Posted by AnthonyPaulO

Hi!

You mentioned having to type in -v for unpaper and I was wondering whether you were only referring to the command-line usage or if I have to start my unpaper switches in the GUI's textbox with -v.

Regards,

Anthony

No, you don't need to specify "-v" if you have any other parameters to pass in that input box.

Just one parameter (simplist one is -v) in the GUI screen input box invokes the unpaper processing within PDFRead i.e. it allows unpaper to work on the images using the defaults.

[AnthonyPaulO]

Hi!

Okay, I've been fooling around with different unpaper options and found something that works for me :

Kindle 1/2
======
out-format : prc
profile : prc-mobi-p
layout-mode : portrait-full
unpaper : --pre-border 0,180,150,0 --no-border-scan
rotation : none
dpi : 300
cropping : 2.0
size : 600x800
colorspace : gray
colors : 16
edge : 5
optimize png's

The things to look at are the unpaper options, specifically "pre-border 0,180,150,0". These are specified in pixels as 0 from the left, 180 from the top, 150 from the right, and 0 from the bottom. Everything outside this rectangular region gets blanked out. Use 0 if there's nothing to remove, in this case I had nothing to remove from the left and bottom of the pages so I left them as 0. However, the top of the page contained headers as well as page numbers, and the right side contained chapter numbers, which is why I stripped 180 from the top and 150 from the right. Now to be technical, pre-border doesn't strip-out or remove the areas, it simply whites them out, but because of the processing that occurs right after it works out to being essentially the same thing. "--no-border-scan" simply tells unpaper that you don't want any borders.

Now each PDF is going to be different, so you have to flip through the pages and check out the margins and come up with your own numbers for pre-border, but that's pretty much all you need to change.

Nick : There's still a small border around the pages no matter what I tried, so I'm wondering whether or not it's even possible to use up all of the Kindle 2's 600x800 resolution. Based on the options I've used, have I missed something or is that pretty much all that can be done? From the samples I've downloaded to my Kindle 2 (including the dictionary and user guide) I think the Kindle has a built-in border that simply can't be used up, so I *think* this is the best that can be done but I want to make sure.

Also, unpaper doesn't take that long at all, just adds a second or two to each page so feel free to use it.

Regards,

Anthony

[nrapallo]

Originally Posted by AnthonyPaulO

Nick : There's still a small border around the pages no matter what I tried, so I'm wondering whether or not it's even possible to use up all of the Kindle 2's 600x800 resolution. Based on the options I've used, have I missed something or is that pretty much all that can be done? From the samples I've downloaded to my Kindle 2 (including the dictionary and user guide) I think the Kindle has a built-in border that simply can't be used up, so I *think* this is the best that can be done but I want to make sure.

Also, unpaper doesn't take that long at all, just adds a second or two to each page so feel free to use it.

Regards,

Anthony

If the small border left over is a white border/margin, then look at this post for an explanation of why this might happen. Nothing you can do about it while preserving the aspect-ratio!

To increase the resulting font size (and improve its readability), you may want to try a landscape profile and landscape-half layout mode. Though, if its large enough for you in portrait mode, then by all means go with that!

Can you post your example .pdf and the resulting .prc (screenshots would be fine)? What was the size of the pdf images in pixels? what percentage would 180 or 150 pixels represent?

If based on dpi and say letter sized .pdfs, then 300x8.5 = 2250 pixels wide, so 150/2550 = 6% or 150/300 = 0.5" from the right and 300x11 = 3300 pixels tall, so 180/3300 = 5.4% or 180/300 = 0.6" from the top.

Does that sound about right?

The PDFRead profile's length and width (Hres & Vres)do represent what was discovered to be the max. resolution of images for the Kindle 1. I would think the Kindle 2 would be the same i.e. max image size would be less than 600x800!

[AnthonyPaulO]

Hi!

Yes, during my tests I had already given thought to the aspect ratio but that's definitely not it since the small border exists all around, not just either vertically or horizontally... had it been the aspect ratio then it would have used up all the remaining space for either the vertical or horizontal edges, see what I mean? It was interesting to see my Kindle 2 go into sleep mode, it displays a picture on the screen and even that picture had a small white border around it so I *believe* that's hard-wired, but I wanted to make sure.

Regarding landscape mode, personally I prefer not to chop the page in half, I want to be able to see the full page all at once, and as long as I can get rid of as much margins as possible I think it's pretty readable for me. Landscape would mean that things would get chopped in half, including sentences.

Regarding the max res, you may be right, I was simply going by their official specs. I can email you the PDF and .prc file if you like... I'll see if I can get your email from your profile when I get out of this editor.

I just tested by removing the 600x800 override, no difference.

But as I'm typing this, something funny just happened... I was converting a full PDF using the settings I posted (654 pages) and something weird happened... the dos process finished and windows explorer opened up to the temp page holding all the png files (is it supposed to do this? How come this never happened before). Being curious I looked through the png's and they are perfect, I mean not a single white border to be found, they are all shaved to the bone. This means something's definitely up, if the png images are perfect then either the process that's packaging these png's into .prc is adding a border or the kindle has a built-in border of its own. Anyway, I close out of this temp folder I realize that the output file isn't there. That's two weird things in a row, I've spent the past 6 hours testing and this hasn't happened before. Fishy!

Also, I've noticed an artifact... on my Kindle reader, there seems to be a blank page inserted here and there in places where there's no blank pages. I checked the generated png's and they're fine, no blank pages there so either it's something in the .prc generation or it's a Kindle bug.

Regards,

Anthony

[nrapallo]

Originally Posted by AnthonyPaulO

Hi!

Yes, during my tests I had already given thought to the aspect ratio but that's definitely not it since the small border exists all around, not just either vertically or horizontally... had it been the aspect ratio then it would have used up all the remaining space for either the vertical or horizontal edges, see what I mean? It was interesting to see my Kindle 2 go into sleep mode, it displays a picture on the screen and even that picture had a small white border around it so I *believe* that's hard-wired, but I wanted to make sure.

It may well be a hard-coded margin or else maybe the max screen size is being displayed and THEN the hard-coded margin kicks in...

Regarding landscape mode, personally I prefer not to chop the page in half, I want to be able to see the full page all at once, and as long as I can get rid of as much margins as possible I think it's pretty readable for me. Landscape would mean that things would get chopped in half, including sentences.

If it works in portrait mode, then go with that. You really don't split sentences in landscape mode since there usually is some overlap between pages to allow you to continue with the non-sliced sentences on the next page.

Regarding the max res, you may be right, I was simply going by their official specs. I can email you the PDF and .prc file if you like... I'll see if I can get your email from your profile when I get out of this editor.

I'll check it out for you once I receive them.

I just tested by removing the 600x800 override, no difference.

But as I'm typing this, something funny just happened... I was converting a full PDF using the settings I posted (654 pages) and something weird happened... the dos process finished and windows explorer opened up to the temp page holding all the png files (is it supposed to do this? How come this never happened before). Being curious I looked through the png's and they are perfect, I mean not a single white border to be found, they are all shaved to the bone. This means something's definitely up, if the png images are perfect then either the process that's packaging these png's into .prc is adding a border or the kindle has a built-in border of its own. Anyway, I close out of this temp folder I realize that the output file isn't there. That's two weird things in a row, I've spent the past 6 hours testing and this hasn't happened before. Fishy!

That happens by design when the creation of the .prc fails. Perhaps it couldn't write to a file that was write-protected or opened by you somehow. Anyways, this can also be forced to occur even with a successful build if you place a file called "debug" in the PDFRead install directory. Good for debugging/investigation purposes!!!

Also, I've noticed an artifact... on my Kindle reader, there seems to be a blank page inserted here and there in places where there's no blank pages. I checked the generated png's and they're fine, no blank pages there so either it's something in the .prc generation or it's a Kindle bug.

Regards,

Anthony

Sometimes, a perfectly normal image created from the pdf page becomes "lost" when converted to .png. I've seen this happen before and don't know what's causing it. Another reason this may happen is when the size of the image is too tall and thus forces a pagebreak, but maybe not. If you can reproduce it with different settings, then it may be more evident which type of error it is. Report back here with some "insights" if you can.

Did you try any other interesting 'unpaper' options?