[sane-devel] Monolta Dimage Scan Dual II (avision backend)

[Alex]

Hi,

I recently acquired a Monolta Dimage Scan Dual II which would not scan 
using SANE (1.2.1 in that case, but the relevant code is identical in 
master). It would just hang when uploading the gamma table, then timeout 
and abort the scan. The attached patch fixes scanning with my particular 
device.

Comparison with the windows driver's USB bus traffic shows that my 
particular unit needs to be sent 4096 bytes of gamma table. It reports 
an ASIC of 3, i.e. AV_ASIC_C2, so SANE 1.2.1 sends it 512 bytes. This 
write times out, and so do all further commands sent to the device 
afterwards. I suspect the firmware ignores the length provided in the 
send command and just expects 4096 bytes anyway? It would sure explain 
the observed behavior since the next 3584 bytes of commands will be 
interpreted as gamma table data instead...

In any case, setting gamma_table_raw_size = gamma_table_size = 4096 in 
send_gamma() works and the device scans beautifully in 8-bit mode. In 
the patch I set that for all AV_ASIC_C2 devices... Not sure if that's 
safe to do, or if it's specific to this particular model and will break 
other models.

12-bit mode didn't work properly (very strong banding). The related code 
in reader_process() is clearly wrong: It's a 12- and 16-bit only 
codepath, it produces a 16-bit sample... but then writes only a single 
byte (lowest 8 bits) to the output buffer. The output buffer clearly 
advances by 2 bytes per pixel, so I just made it write a 16-bit 
host-endian pixel instead. Scanning then works perfectly in 12-bit mode 
as well. That change should also fix 12- and 16-bit scanning for all 
other avision devices (I don't see how any of them can scan correctly in 
12- or 16-bit mode when only 8 bits of the sample data are written).

Please let me know if you need further information or need me to test 
something.

Regards,
- Alex

PS: Sorry for reporting this the old-fashioned way. I don't have an 
existing gitlab.com account, and their current verification process is 
quite excessive IMHO.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: avision.patch
Type: text/x-patch
Size: 619 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/sane-devel/attachments/20260805/1a468f7f/attachment.bin>