[sane-devel] problems with scanimage to pdf

[Simon Matter]

> On 02/04/2015 04:17 AM, Carsten Jensen wrote:
>> On 02/03/2015 10:58 PM, m. allan noah wrote:
>>> Most scanners produce big-endian data, which scanimage writes out to
>>> disk when it creates the tiff. Any host-side tools that you use on the
>>> tiffs will probably convert them to little-endian, since that is the
>>> format of your CPU. If you attempt to rotate all of them (some of them
>>> by 360 degrees), such that they all become little endian, and all get
>>> the strip 204 error fixed, does the pdf work better?
>>>
>>> allan
>>>
>>> On Tue, Feb 3, 2015 at 4:43 PM, ken <gebser at mousecar.com> wrote:
>>>> On 02/03/2015 03:21 PM, Carsten Jensen wrote:
>>
>> If the above doesn't work you could -as a test- make all the tiffs into
>> pdf before manipulating them.
>>
>> Basically you'll need to find where the error occurs.
>>
>> I hardly doubt this is a bug from sane.
>> but upgrading your imagemagick might do some good.
>>
>> Carsten
>>
>
> Thanks to all for the suggestions.  I've been trying and testing all
> kinds of different things for a bit over two full days, much longer than
> something like this should take.  Last night I tried hp-scan, the
> utility that came from HP with their RPM package.  It produced a really
> nice pdf on the first attempt... except, as mentioned a few days ago,
> the backside of every physical page (sheet of paper) was upside-down.
> This morning I figured out how to rectify that with pdftk.  So now I'm
> getting the pdf document I was after in the first place.  I have to say
> that the end-product is great-- so clean and clear (MUCH better than I
> was getting with SANE) at 300dpi that I'm going to try 200dpi.  Then I
> can script it, knowing that all the commands actually work as advertised.
>
> I've been a FOSS junkie since '92 and would have preferred staying with
> it (as much as is possible under the circumstances).  But at some point
> I have to get back to the rest of my life.  At least pdftk is open
> source.  And I'll probably use SANE in future for some less demanding
> projects.

I may be wrong but I always thought hp-scan is also open source and is
simply using SANE to do all the real work, as written here
http://hplipopensource.com/hplip-web/tech_docs/man_pages/scan.html
So you don't have to wait for less demanding projects to use it :)

The real question is why your results are so different with different
front ends on the same SANE installation?

Regards,
Simon