[sane-devel] terrible (systematically corrupted) jpg at 1200dpi

ken gebser at mousecar.com
Fri Jan 30 09:43:54 UTC 2015

The apps I've been using to examine the scans I've made so far are 
ImageMagick's "display" and "gimp" Both of these-- which I've used 
countless times over many years-- always worked fine in the past. But 
for some reason, the images scanned at 1200 dpi looked horrible when 
displayed by either of them (reduced visually to 3 or 4% of the original).

So I looked at the attachment uploaded to the hplip site, and a couple 
other of the images I scanned, using Firefox and-- happy surprise-- they 
looked fine: There was no distortion, no bands of discoloration, and the 
images were clear and focused... completely unlike what either "display" 
or "gimp" rendered for me.

I don't recall ever having any problems using "gimp" or "display" to 
look at jpg images before, even files well over 100M. With my previous 
Linux PC and scanner I did 4800x4800 scans-- I could see individual 
fibers in the paper-- and didn't have any problems then. So it looks 
like I need to investigate what's going on with "display" and "gimp". 
But that's of no concern to this list. So we can count this issue Closed.

On 01/28/2015 02:09 PM, Carsten Jensen wrote:
> the circles with the dot in the middle that you're refering to in one of
> your initial email
> are the result of the print, an algorithm to adjust the brightness of
> colours, and make the print cheap.
> you can see them if you use a magnifying glass.
> you see them now because of the high resolution and you can zoom in.
> I must agree with Allan, I don't see anything wrong with the images either.
> Carsten
> On 28-01-2015 19:58, m. allan noah wrote:
>> I just looked at the top couple inches of both images in your
>> launchpad bug report, and I dont see anything wrong with the files.
>> allan
>> On Wed, Jan 28, 2015 at 1:53 PM, ken <gebser at mousecar.com> wrote:
>>> I already opened up a bug report on this:
>>> https://bugs.launchpad.net/hplip/+bug/1415121, if anyone's interested.
>>> On 01/28/2015 01:44 PM, ken wrote:
>>>> What I wrote already in the prior email makes it clear this is a 1200dpi
>>>> scanner.
>>>> "4 pixels per dot" would correspond to 3 colors + a B/W bit for each
>>>> dot-- what would be required for color scanning at 1200dpi.  "Pixel" is
>>>> a rather slippery term.  Perhaps the tech should have said "sensel". But
>>>> this was tier-one tech support after all, not a developer I was talking
>>>> with.  And she had to look up that much.
>>>> On 01/28/2015 10:33 AM, m. allan noah wrote:
>>>>> "4 pixels per dot" is a meaningless statement. A dot is a pixel. You
>>>>> would have to ask the hplip guys if they can get 1200 dpi out of this
>>>>> scanner.
>>>>> allan
>>>>> On Wed, Jan 28, 2015 at 9:39 AM, ken <gebser at mousecar.com> wrote:
>>>>>> On 01/28/2015 07:23 AM, m. allan noah wrote:
>>>>>>> at least in lineart mode, the scanner maximum is 300dpi. Try scanimage
>>>>>>> --mode=color --help
>>>>>>> allan
>>>>>> According to specifications published by HP, this scanner's hardware and
>>>>>> optical resolution is 1200x1200 dpi.  I just called HP tech support to
>>>>>> confirm this and was told this is correct and that, further, there
>>>>>> are "4
>>>>>> pixels per dot", meaning that the scanner handles both color and B/W
>>>>>> at that
>>>>>> resolution.  So I don't know why scanimage would report that 300dpi
>>>>>> is the
>>>>>> maximum resolution.

More information about the sane-devel mailing list