[sane-devel] PIXMA: MX360 at 2400 x 1200 dpi

[Rolf Bensch]

Am 18.02.2012 12:40, schrieb ibu ☉ radempa ䷰:
>>
>> You need to investigate 2 things:
>> (1) 1200 x 1200 dpi:
>> (a) Does the scaned image contain all expected pixels (w x h)?
>
> Yes, the image size is correct.
>
>> (b) Please check the pnm image for mixed columns. It seems that two half
>> pictures are bitwise mixed. The result is a streched image with a shadow
>> of the 2nd half picture.  It's easy to solve this line by line with a
>> (new) reorder_pixels function. You can have a look to similar functions
>> in the pixma_mp810.c file.
>
> With the attached patch (just disabling pixel reordering, as for MP220)
> everything ist fine, please commit it.

OK

>
> (To exclude yet another pixel reordering mistake I tried to swap even
> and odd pixels (
>
> memcpy (linebuf + c * (1 - (i % 2) + (i / 2) * 2), sptr + c * i, c);
>
> ) and got parity_inverted.pnm while without reordering I got
> correct_order.pnm .)
>
>> (2) 2400 x 1200 dpi:
>> Here we should take care that the scanner gets the correct formatted
>> scan command.
>> Can you sniff the usb protocols on a windows machine and compare it with
>> one from a linux machine? Both for 2400 x 1200 dpi. The positions of the
>> parameters can be read from the send_scan_param function. You can use
>> e.g. wireshark for sniffing. The scan command starts with "d8 20 00 00
>> 00 00 00 00 00 00 00 00 00 00 00 38".
>
> Luckily I don't have windows, so I tried Canon's scangearmp 1.70-1
> package for debian and found that it also allows only up to 1200 dpi. I
> suggest to leave 1200 dpi as maximum resolution. (For completeness I
> attach some hexdumps of the respective usb request blocks.)
>

OK, we leave 1200 x 1200 dpi as maximum resolution.

> In the list of working devices on the website I'd suggest to add a
> comment for Canon MX360: ADF working partially.
>

In sane-pixma.man I added MX360 to the list of sometimes hanging 
scanners and in pixma.desc I added the comment: ADF empty buggy.

I just committed everything to git.

Many thanks for reporting and debugging this scanner.

Cheers,
Rolf