Page 27 of 28

Re: Scan Tailor "Experimental"

Posted: 10 Oct 2016, 08:23
by d14b0ll0s
Tulon wrote:At the peak of donations I was receiving something equivalent of 3 days of my day job salary in a year of donations.
At the point when I stopped accepting donations, I was receiving something equivalent of 2 hours of my day job salary, again in a year of donations.
Sorry to hear that. I hope there will be something to convince you to work on this further in the future, however; perhaps it would be easier to manage as a collaborative project?

Re: Scan Tailor "Experimental"

Posted: 10 Oct 2016, 19:29
by Tulon
d14b0ll0s wrote:perhaps it would be easier to manage as a collaborative project?
It's already managed like that. Well, not the Experimental branch but the main one. There are people who respond to bug reports and can accept a patch, yet hardly any development is going on.

The thing is, I am not the only one who doesn't want to work on the boring stuff for free. In fact, when it comes to working on someone else's code, there is even less of incentive. Almost all of my contributions to opensource projects other than my pet ones were made as part of my employment. The only significant 3rd party contribution to Scan Tailor, namely the command-line mode was made as part of someone else's employment.

Re: Scan Tailor "Experimental"

Posted: 11 Oct 2016, 03:56
by d14b0ll0s
I see the issue. But do you think the experimental branch could be managed similarly to the main one?

Re: Scan Tailor "Experimental"

Posted: 14 Oct 2016, 04:52
by Tulon
d14b0ll0s wrote:I see the issue. But do you think the experimental branch could be managed similarly to the main one?
Maybe it could, if there is enough interest from the community. But then, what's the point? Right now, when you report a bug or submit a feature request on the main branch, you get a polite response acknowledging it, but there is hardly a chance of anything being done about it. This situation must be quite frustrating both for bug reporters and for people responding to them.

Re: Scan Tailor "Experimental"

Posted: 10 Mar 2017, 20:14
by bgd
I know that Scan Tailor is not under active development but, either way, here are two features that I'd love to see in a potential future update:

Auto-enclose picture zones: The automatic picture zone selection sometimes leaves cavities within a picture area and this feature would avoid that by simply drawing straight enclosing lines between the outermost points which would enclose the complete picture area within a polygon-shaped area (usually a rectangle but could also result in a triangle, pentagon, hexagon, etc, or any random polygonial shape). This option would in other words take a more aggressive approach when auto-selecting picture zones and would thus avoid potential cavities. From my experience, more is almost always better when defining pictuire zones in Scan Tailor.

Manual selection of deskewing zones: Often when dealing with severly and irregularly curved lines, Scan Tailor (by design of its algorithm) skews one end of the page when deskewinng the other end. This is the result of overcompensation by the algorithm and probably the easiest way to fix this (easier than creating a new improved algorithm from scratch) would be to let the user define one or more deskewing zones on the page. The simpler variant of this would be to let the user define one single active deskewing zone on the page and Scan Tailor would then ignore the rest of the page when deskewing. A more advanced variant would let the user select several standalone active deskewing zones.

Re: Scan Tailor "Experimental"

Posted: 28 Aug 2017, 14:27
by tomse
When I first found scantailor -introduced to me here by this site, thanks- back in 2013-14 I believe it was, there was no doubt that this awesome piece of software reduced my time spent creating my manuals. I did donate back then, but was sorry to see that development had stopped.
It's good to see that you're working on it again from time to time.
I know this thread is old, but it seems there's still life in here :-)

I found some inconveniences in the newer version, and I have some feature requests, should you ever want to take them up :-)
I'll gladly send a donation your way (again) regardless as you've actually kept my work going. If you have a paypal account please send me a PM with it then.

my feature request is as follows

Mark the talest and widest page
and/or mark page for max range.
* sometimes pages get modified using margin, contents or whatever, and sometimes the content pages are larger than the frontpage, which should be the largest in both height/width, so adding a colour or whatnot to the largest and one for the widest would help a lot looking for those pages, sometimes multiple pages are larger than the cover so the list needs looking through several times.
a Max limit could be useful

min size of contents so pagenumbers don't get excluded
Most of the times I have to manually adjust the content box as the automatic ignores page numbers when they're too far from the contents.
Setting a minimum size box could be useful here

ignore an area when doing autocontent box
scanning books with spiral spines, creates a row of holes, which are black when scanning B/W.
these holes gets added to the automatic content box. it would be really nice to exclude such an area before
running that part. of course it should apply to every other page as odd pages have the holes in the left side, and even
pages have them in the right side.

fix black'n'white cropping
when outputting inverted text, chapter pages, what is supposed to be a black box gets "equalized" even though that feature isn't checked

export to G4 compression rather than LZW for B/W
this is rather self explanatory. everything gets output as LZW, but using G4 fax compression / CCIT.6 or what it's called would reduce filesize to a 1/3.

margin = mm rather than %
the new % is somewhat confusing, and I miss the old scale in mm.

output is in 96dpi
output is in 96dpi, and you cannot choose a custom resolution as you used to. please bring back the different dpi settings again.

fill zones, picture zones, resolution etc could be moved to a step between margin and output so one wouldn't have to wait for an image to be output before you can change these settings.

crash - (maybe fixed?)
I don't know if you've fixed this, but there used to be a nasty bug which made the application crash hard when outputting pages. It mostly happens whith manually content adjusted pages. The workaround was to adjust the selection a single pixel.
If it happens again, I'll save the whole project for you to test.

once again, thanks for making my work a lot easier :-)

Re: Scan Tailor "Experimental"

Posted: 31 Aug 2017, 04:19
by Tulon
Hi tomse,

It's been a while since I worked on Scan Tailor Experimental or any other version of Scan Tailor. There is also no one else who could implement new features. I can only suggest you to make sure you are running the latest version, which you will find at the top of this page.

As for 96 DPI output images, take a look at my signature.

Re: Scan Tailor "Experimental"

Posted: 29 Sep 2017, 13:34
by tomse
Thanks for getting back to me.
it _is_ the latest experimental version I'm running, since I recently just found it :-) (btw a version number would be nice in help->about)
I was waiting for a mail notification, but apparently forgot to check that here in the forum :-)

the 96dpi problem didn't exist in prior original release nor the enhanced version, hence I felt like mentioning it.
at present time I'm coding an app that will help me speed up the process for scantailor pre and post work.
I've been running a lot of scripts to do this tedious work (contrast fixing, compress to G4 fax etc and also resample to correct resolution),
which now I'm implementing to a nice gui app, well actually it's just a wrapper for imagemagick/pdftk.

I can't code all the fancy gfx stuff nor in c++ so this will have to do, but you actually inspired me to make it multithreaded seeing as ST now do this, which is so awesome!.

Re: Scan Tailor "Experimental"

Posted: 04 Oct 2017, 10:19
by Konos93a
hello sir

u have done a great work . if scantailor had "crop" , before "split pages"
"straighten text lines" like abbyfinereader before "select content" or in geometric distortions first the command "page skew" and after "keystonig" then i would stop to use abbyfinereader and use only scantailor to edit images .

here in greece i try to enlarge scanning population .

anyway the request is other . i take photos in large resolution (2523*4180) 210 kb and this in command
"output -resolution enhanchemen 1x" so this book 43.8 mb in pdf and that how i upload it

the problem now is that in my kobo aura one all this resolution is not nececarry ( i run there koreader) and stuck so i want a command "output -resolution enhanchemen 0,5x" to have an image 1261 * 2090 there where it write (1,5x) .

how can i change this?
how difficult is to chose a output -resolution enhanchemen from 0,5 till 5 in a scale ?
what type of scanner do u use ,what camera do u have , how do u trigger them ?

Re: Scan Tailor "Experimental"

Posted: 29 Apr 2018, 06:50
by tomse
Having used Scantailor experimental extensively for some time now, I've build myself a little library of scripts that I need to run every time I've finished off with Scantailor some of them are fixing dpi, compressing using G4, convert 16bit page to grayscale etc.

In the previous versions ST used to crash and I'd have to adjust content box on those pages. it's AWESOME that this has been fixed in this version.
I'm not using any of the older versions as this is the best one available (even due to the feature wishes below).

The DPI problem - which most applications really don't like (slow downs etc), i.e. pdf applications. I'm fixing using imagemagick and resample.
(it would be REALLY nice to get added that ST applies this)

for Black/white output, G4 fax compression would be great to have this added as well, 1/3 of the size of LZW still lossless.

using % for margins, I really can't understand why it was changed to this, pages with different sizes can't have adjusted i.e. left margin to 1 value, and applied to the rest of the pages without having to work on every page to get a proper alignment.

the logick doesn't always take the whole contents, it can be due to the page number in the bottom is too far away from the text or page number/top is separated by a line. It would be real productive to be able to have a type of content box where you approximate that everything within here should be added to the automatic content selection. this would also be useful to exclude things, like holes from a ring spine etc.

one thing I really miss from the Enhanced version is the ability to select "this page and the following" rather than just all pages.

Tulon, is there any way of motivating you to do any of the requests?