[sane-devel] artec e+pro- vertical lines- some samples

[kilgota at banach.math.auburn.edu]

On Sat, 12 Jan 2008, Philip Aston wrote:

> Yes I agree it was REALLY stupid but the penny dropped (about the black
> hat and the spam or worse) immediately after that email, so I changed
> the password, before sending the "retraction" in (my next email), thats
> why no-one could access the account.  When I logged back into it I was
> extremely fortunate there had been no activity! I fully accept that msn
> require us not to do this for a very good reason and I hope thats the
> last major senior moment I experience.
>
> The files are now on http://homepage.ntlworld.com/philip.aston/ and I
> apologise for that ridiculous mistake!


Philip,

Indeed, this seemed like a very novel way to post data. So I decided that 
I should wait until the dust settles, which it now seems that it did.


What I see in these images is, something like what I suspected could be a 
possibility, or it could be something else. It is difficult to tell 
without the actual raw data. I assume (without actually knowing) that 
there is a SANE option to capture the raw data and dump to a file, with no 
processing at all.

Remarks:

1. I cannot get actively involved in this. I just saw something in what 
you posted in the first place which caught my eye and looked like a 
problem which I have seen before, several times. The images that you 
posted confirm that I may -- or may not -- have been correct in my 
suspicions. But I am not going to get involved right now in supporting 
scanners. Moreover, you are dealing with a SANE developer right now, and 
it is both my inclination and my duty not to get in between. That said:

2. Possible reasons for the problems, with examples:
 	-- reversal of order with some rows (libgphoto2/camlibs/jl2005a)
 	-- reversal of order with some columns (libgphoto2/camlibs/aox)
 	-- reversal of column-pairs needed after decompression even though 
Bayer tiling is "correct" (libgphoto2/camlibs/sonix)
 	-- other reasons of which I know nothing because I am not doing 
scanners, know nothing at all about the internals of this particular 
scanner driver, and do not have the source code for anything near a 
current version of SANE.

3. In order to work with something like this, one needs the raw output, 
preferably after decompression if the raw output from the scanner had to 
go through a decompression stage. If someone wants to send me this or 
post it, I will be glad to play with said raw data and see if the output 
can easily be tweaked to look right.

4. Recently, I posted one of my tools for fooling with raw output to the 
SVN tree of gphoto, at gphoto.svn.sourceforge.net, under 
trunk/playground/raw_converters. The program will do the following for 
photos from a camera with JL2005A chipset:

-- convert the raw data for a photo, obtained with something like
 	gphoto2 --get-all-raw-data
into a PPM image file.
-- display said image in an xterm window
-- exit when the window is closed, and save the output file.

This is for a specific camera's output, of course, but the program is at 
least semi-modular and it is fairly easy to adapt to other cameras (which 
I have done several times, for my own use). It should be fairly easy to 
adapt to handle raw output from a scanner, too. The only problem would be 
to have some way to put in the width and height of an image.

The reason why I wrote this little program is that it makes it possible to 
process the same data repeatedly, while introducing small changes in the 
processing routine along the way in order to improve things. That way, one 
can be more assured of getting things right, in the end. If anyone would 
like to use it here, then welcome to it. If I could have the raw data 
_and_ the currently possible finished product, I could also see what I 
could get out of the raw data. I would be glad to do that, and I think I 
would have time to do that. If in fact the raw data from the scanner is 
compressed, I would of course need to be pointed at the decompression 
algorithm, which is to be found somewhere in the code for the scanner 
which is to be found somewhere in the SANE source tree.

Theodore Kilgore