#389 Support sane option --scan-area

[CoSoCo]

Several sane backends provide the option --scan-area to guarantee the scan is correctly aligned, even if the scan area with automatic document feeder (ADF) is right-aligned in contrast to the flatbed of the same device, which is normally left-aligned.
Currently on Ubuntu gscan2pdf seems only to support options -l -t -x -y, which leads to misalignment of some scanners when using the ADF.
As example for this problem see: https://gitlab.com/utsushi/utsushi/issues/68

[Jeffrey Ratcliffe]

Please start gscan2pdf from the command line with the --log=log options, reproduce the problem, quit, and post the log file, which gscan2pdf should have compressed with xz.

[CoSoCo]

Here it is ...

[Jeffrey Ratcliffe]

Thanks for the log file. It looks to me as if you are setting scan-area, as well as br-y, br-x, and tl-x.

If you delete or rename ~/.config/gscan2pdfrc, start gscan2pdf, and then only set source=ADF and scan-area=A4, it the scan still misaligned?

[CoSoCo]

With cleaned config: This is what I get, if I first set ADF and then scan area to A4

[CoSoCo]

With cleaned config: This is what I get, if I first set scan area to A4 and then ADF

[CoSoCo]

Here an example, where I don't remember which settings I've used in which order, but this has worked fine except that the right-hand border is 1 mm too large.

I too have the problem, that I can't scroll in the image view. I can zoom with the mouse wheel, but there are no sliders to scroll. I don't know what the pan tool is for, but the button is always pushed. I can not unselect it by mouse click.

What is the difference between scan area, paper size and quick format?

[Jeffrey Ratcliffe]

Can you give me another log file, please, following what you did for one of the two TIFs?

Additionally, what do you get from the following from the command line:

scanimage --source='Automatic Document Feeder' --scan-area=A4 > out.pnm

?

[CoSoCo]

Your 1st question I don not understand. Can you explain more detailed?

Here is what I get from scanimage.
Here also the right-hand border is 1 mm too large.

Is there a way to directly compress a tif file with G4 with scanimage to get a much smaller file?

[Jeffrey Ratcliffe]

I'm guessing that scanimage -o out.tif will guess that the output should be TIF. Alternatively, --format should work, too.

I meant before that the first two TIFs looked too tall, and now that you have cleaned the config file, I'd like to understand from a fresh log file what is going on. So please start gscan2pdf, and then only set source=ADF and scan-area=A4, quit and post the new log file.

[CoSoCo]

I'm guessing that scanimage -o out.tifwill guess that the output should be TIF. Alternatively, --format should work, too.

I know this options, but AFAIK they don't allow to set a compression. But I have found this workaround:
scanimage --source='Automatic Document Feeder' --scan-area=A4 | convert - -compress Group4 "scanimage area_A4 G4.tif"
Surprisingly this caused a file size of 230 kb, whereas gscan2pdf only needed 100 kb in G4 mode. Is there a preprocessing in gscan2pdf, which optimizes the image, e.g. from noise?

I meant before that the first two TIFs looked too tall, and now that you have cleaned the config file, I'd like to understand from a fresh log file what is going on. So please start gscan2pdf, and then only set source=ADF and scan-area=A4, quit and post the new log file.

This is exactly what I did in the first 2 examples, with alternating order of setting the 2 options. It is surprising to me too, that anyway maximum size was processed. I have no idea, in which way I may have misunderstood you.

Only in the 3rd example I got the right result. I don't remember the exact order of options I used, but I hoped you could see this from the log file.

[Jeffrey Ratcliffe]

Group3 and 4 only works for bilevel - black & white, not greyscale - images.

I'm guessing that the default scan mode is greyscale, which Group4 can't compress, so convert is probably just ignoring it and using something like Flate instead. What does tiffinfo say about the file?

I don't see a log file after you cleaned your config. Where did you put it?

[CoSoCo]

Oh god, now I understand. There must have happened an error when uploading 2 files at a time. Now I have uploaded the 3 missing log files. Sorry for my overlooking.

The result with scanimage + imagemagick convert:

$ ls -l scanimage\ area_A4_G4.tif
-rw-rw-r-- 1 ich ich 263640 Jul  5 20:03 'scanimage area_A4_G4.tif'
$ tiffinfo scanimage\ area_A4_G4.tif
TIFF Directory at offset 0x40516 (263446)
  Image Width: 2480 Image Length: 3507
  Bits/Sample: 1
  Compression Scheme: CCITT Group 4
  Photometric Interpretation: min-is-white
  FillOrder: msb-to-lsb
  Orientation: row 0 top, col 0 lhs
  Samples/Pixel: 1
  Rows/Strip: 3507
  Planar Configuration: single image plane
  Page Number: 0-1
  ImageDescription:  SANE data follows

The result with gscan2pdf:

$ ls -l gscan2pdf-test\ ADF.tif
-rw-rw-r-- 1 ich ich 99804 Jul  5 17:36 'gscan2pdf-test ADF.tif'
$ tiffinfo gscan2pdf-test\ ADF.tif
TIFF Directory at offset 0x1848e (99470)
  Image Width: 2480 Image Length: 3507
  Resolution: 300, 300 pixels/inch
  Bits/Sample: 1
  Compression Scheme: CCITT Group 4
  Photometric Interpretation: min-is-black
  FillOrder: msb-to-lsb
  Orientation: row 0 top, col 0 lhs
  Samples/Pixel: 1
  Rows/Strip: 3507
  Planar Configuration: single image plane
  Page Number: 0-1
  ImageDescription:  SANE data follows

  White Point: 0.3127-0.329
  PrimaryChromaticities: 0.640000,0.330000,0.300000,0.600000,0.150000,0.060000

Maybe the threshold for black/white is different between gscan2pdf and imagemagick. Also note the difference with the Photometric Interpretation value.

[CoSoCo]

I too have the problem, that I can't scroll in the image view. I can zoom with the mouse wheel, but there are no sliders to scroll. I don't know what the pan tool is for, but the button is always pushed. I can not toggle it by mouse click.

And can you tell me what is the difference between scan area, paper size and quick format?

[Jeffrey Ratcliffe]

To use the pan tool, hold the left mouse button down in the image view and drag. If the screen is split so the text is also visible, the text should be panned with the image.

scan area and quick format are options offered by the backend. paper size is added by gscan2pdf for those backends (like mine) that don't have scan area.

[Jeffrey Ratcliffe]

I expect you are right. The different black/white and threshold settings meant that the image from gscan2pdf had much more white in it, which group4 could compress much better.

[CoSoCo]

To use the pan tool, hold the left mouse button down in the image view and drag.

This does not work. In the image view I see a cross-sized cursor, and when I drag it, an area becomes selected. The pan tool button is selected from start and I cannot toggle it.
Also the text edit button is selected from start, and I can't unselect it.

But when I choose"pan & select", the pan tool button becomes unselected. If I select it again,, I see a finger-pointing cursor and I can move the image with dragging the mouse.

I have no idea, what the "pan & select"-function is good for.

[Jeffrey Ratcliffe]

I've just seen the following bug in the image controls, which is probably what is confusing you: The controls default to the "pan & select" tool, but the icon defaults to the pan tool. I'll fix this.

It should work as follows:

• pan & select: LMB select, scroll wheel zooms, MMB pans.
• pan: LMB pans, scroll wheel zooms
• select: LMB selects

[CoSoCo]

Don't forget, that the text edit button has the same problem.

[Jeffrey Ratcliffe]

Don't forget, that the text edit button has the same problem.

What do you mean by this? Are you talking about panning and scrolling the text (i.e. OCR output) window?

[CoSoCo]

I would like a more fine grained zoom, e.g. like in IrfanView. There it is:
1;2;3;10;14;16;19;21;25;28;33;38;43;50;57;66;76;87;100;115;132;150;174;200;230;260;300;350;400;450;500;600;700;800;900;1000;1200;1400;1600;2000;3000;
I have a very fine grained mouse wheel and an older CPU, so when I turn the mouse wheel just a little, I first have no reaction and then a biiig jump after few seconds. So it is very uncomfortable to zoom with the mouse wheel here.

But what about the main problem about the wrong alignment of ADF scans with my Epson WF-2630 ???

In the case of this comment, I remember, that the left edge value (l) was automatically set to 6 mm (exactly it should be 5.9 mm), the width (x) to 210 mm and the height (y) to 297 mm, which is the correct setting for right-aligning ADF scanners such as my Epson WF-2630. Unfortunately I don't remember the order of actions I did for this.

Anyway, I don't think, it is the right way to change the y,y,w,h settings upon switching to ADF.
Imagine, one would like his scan cropped by 20 mm, then in case of ADF he should enter 25.9 mm.
Better make use of the --scan-area option.

[CoSoCo]

Note, that I've edited my last post.

[CoSoCo]

For comparative examples with scanimage see here: https://gitlab.com/sane-project/frontend/xsane/-/issues/1#note_621030059

[Jeffrey Ratcliffe]

In both gscan2pdf ADF_A4.log.xz (sourceforge.net) and gscan2pdf A4_ADF.log.xz (sourceforge.net), first scan-area=A4, and then source=ADF was selected. It looks as though the ADF overwrites the geometry settings.

It is a bug in the backend, that ADF overwrites the geometry settings without unsetting the scan-area option, or setting it to max, or something like that.

Would you mind please having another go and setting first the ADF, and second the scan-area, and posting the log file and image?

[CoSoCo]

Oh, it seems, that I was in error. Here are the new files.

Additionally I've filed a bug report: https://gitlab.com/utsushi/utsushi/-/issues/68#note_621557725
In the mean time please create a workaround in gscan2pdf.

[Jeffrey Ratcliffe]

The bug is not that the order of arguments is important - but that having first set A4, and then ADF, that scan-area is not set to max - i.e.

scanimage --scan-area=A4 --source='Automatic Document Feeder' --help

still lists scan-area as A4, not max.