Epson 3490

[Jim MacLeod-2]

I'm new to scanning and relatively new to Linux. Searched the web for info
and got the impression, which others confirmed, that generally anything
epson works with Linux, but maybe not the 3490! sane-find-scanner gets
vendor and product code but scanimage -L finds nothing.

Having now read a previous mail about the 4490 I'm guessing that the same
applies to the 3490, ie no go..are my fears justified?

Vuescan which does support the 3490 but I've got to learn more about usb
first before going down that road - or is there hope.
Jim


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Henning Meier-Geinitz]

Hi,

On Wed, Aug 31, 2005 at 04:16:46PM +0100, Jim MacLeod wrote:
> I'm new to scanning and relatively new to Linux. Searched the web for info
> and got the impression, which others confirmed, that generally anything
> epson works with Linux, but maybe not the 3490!

Generally, to find out if a scanner is supported by SANE, enter it
into the scanner search engine:
http://www.sane-project.org/cgi-bin/driver.pl?manu=epson&model=3490&bus=usb

It says that that scanner is supported with support level "basic" by
the snapscan backend.

> sane-find-scanner gets vendor and product code but scanimage -L finds
> nothing.

Maybe your sane-backends package is just too old? As far as I know,
you need version 1.0.16 for that scanner.

> Having now read a previous mail about the 4490 I'm guessing that the same
> applies to the 3490, ie no go..are my fears justified?

That scanner isn't even in our list. Can anyone provide more details
(vendor/product ids, chipset)? Support is planned for the epkowa
backend, see the list archive of this list for details.

Bye,
  Henning

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Jim MacLeod writes:

> I'm new to scanning and relatively new to Linux. Searched the web for info
> and got the impression, which others confirmed, that generally anything
> epson works with Linux, but maybe not the 3490! sane-find-scanner gets
> vendor and product code but scanimage -L finds nothing.

> Having now read a previous mail about the 4490 I'm guessing that the same
> applies to the 3490, ie no go..are my fears justified?

Somewhat.  See the thread starting at

        Message-ID: <m37jet91od.fsf@...>

for my problems - it uses the snapscan backend.  I had to install the
scanner under Windows first, then copy the firmware file to my Linux
partition and edit the config file for the snapscan backend accordingly.

That made scanimage recognize and use the scanner, but it is unable to
scan the full page.  It only scans 75-80% of the width and height.
Subsequent questions have gone unanswered and I've had limited time to
look into the backend code myself so I think my solution is to return the
scanner within the next few days and rather get one that is supported by
SANE.

Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Oliver Schwartz]

Hi,

> It says that that scanner is supported with support level "basic"
> by the snapscan backend.
>
> > sane-find-scanner gets vendor and product code but scanimage -L
> > finds nothing.
>
> Maybe your sane-backends package is just too old? As far as I know,
> you need version 1.0.16 for that scanner.

You'll need the current CVS version for this scanner. 1.0.16 won't
work at all (and see Bjorn's mail for the problems that still exist.
I haven't yet found the time to look into it.)


/Oliver


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Oliver Schwartz]

Hi,

> That made scanimage recognize and use the scanner, but it is unable
> to scan the full page.  It only scans 75-80% of the width and
> height. Subsequent questions have gone unanswered and I've had
> limited time to look into the backend code myself so I think my
> solution is to return the scanner within the next few days and
> rather get one that is supported by SANE.

The guy who sent me the patch for the 3490 had the same problem when
using low resolutions for previews. He reported that the problem
disappeared after the lowest possible resolution was set to 200. The
version I checked into CVS should already use 200 dpi as lowest
possible values.

Can you please check a few things:
- What frontend are you using?
- Does "scanimage" as frontend scan the correct area? E.g. does
"scanimage -x 10 -y 10" scan an area of 10x10 cm? Please try with
different resolutions.
- What firmware file version are you using? The easiest way to find
out is running

strings firmware.bin | tail

on your firmware file.

Unfortunately I won't have much time to look into this in the next
weeks.

/Oliver


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Oliver Schwartz writes:

[...]

> Can you please check a few things:
> - What frontend are you using?

scanimage, the CVS version from 20050815.

> - Does "scanimage" as frontend scan the correct area? E.g. does
> "scanimage -x 10 -y 10" scan an area of 10x10 cm? Please try with
> different resolutions.

I've tried the full area with resolutions from 150 up to 1200, but got
the same reduced (75-80%) area every time.  I've tried a smaller area
with resolution 150, and saw similar reduction in the expected scanned
area, but I'll have to do some more testing tonight to give you
definitive numbers for that.

> - What firmware file version are you using? The easiest way to find
> out is running

> strings firmware.bin | tail

> on your firmware file.

I don't have access to that information right now, so I'll check that
tonight also.  It is whatever firmware came with the scanner -
esfw52.bin.

My main interest at this point is to understand whether this issue is
something that can be solved with some patch to the backend, or whether
it's inherently related to the scanner's firmware or something that is
Epson-proprietary that the SANE developers won't get access to.  Is
anyone able to shed some light on this issue?  It will help me decide
whether to return the scanner as hopeless or hang on to it hoping that a
working patch for the scanner will be available in the near future.

(There are some other issues too like controlling brightness etc. from
the command line doesn't work - e.g. scanning in Lineart mode turns out
too dark compared to what I'm used to from my UMAX Astra 1200 scanner
whose SCSI-interface seems to have broken down.)

Thank you,

Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Oliver Schwartz]

Hi,

> I don't have access to that information right now, so I'll check
> that tonight also.  It is whatever firmware came with the scanner -
> esfw52.bin.

yes, but at least for other epson scanners these firmware files are
available in different versions. The filename is always the same.

Another thing that I'm interested in is whether only the size of the
scan area is wrong or if the origins of the scan area are also not
set correctly (i.e. options -l and -t in scanimage)

If it is always the same ratio of real distance to scanned distance
(also for -l and -t) it is most likely that the calculation of the
scan area is not handled correctly. That could be corrected by
changing the value of pos_factor in function set_window() in
snapscan-scsi.c. For the guy that developed the patch that didn't
help, though.

> My main interest at this point is to understand whether this issue
> is something that can be solved with some patch to the backend, or
> whether it's inherently related to the scanner's firmware or
> something that is Epson-proprietary that the SANE developers won't
> get access to. Is anyone able to shed some light on this issue?  

I don't think that this is an unsolvable problem. It may be necessary
to do some deeper analysis and maybe get some USB traces from the
windows driver, though.

Also keep in mind that the scanner seems to work for at least one guy,
so the solution can't be that difficult :-)

/Oliver


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Olaf Meeuwissen]

Jim MacLeod <j75trym@...> writes:

> I'm new to scanning and relatively new to Linux. Searched the web
> for info and got the impression, which others confirmed, that
> generally anything epson works with Linux, but maybe not the 3490!
> sane-find-scanner gets vendor and product code but scanimage -L
> finds nothing.
>
> Having now read a previous mail about the 4490 I'm guessing that the
> same applies to the 3490, ie no go..are my fears justified?

The same does not apply.  The 4490 and 3490 are very different
scanners.  The 3490 is partially supported by the latest snapscan
backend.  Also, iscan will support the 3490 with the next version
(1.17.0) using a closed source plugin.

> Vuescan which does support the 3490 but I've got to learn more about
> usb first before going down that road - or is there hope.

Hope this helps,
--
Olaf Meeuwissen                          EPSON AVASYS Corporation, LAN
FSF Associate Member #1962           sign up at http://member.fsf.org/
GnuPG key: 6BE37D90/AB6B 0D1F 99E7 1BF5 EB97  976A 16C7 F27D 6BE3 7D90
Penguin's lib!       -- I hack, therefore I am --               LPIC-2

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Olaf Meeuwissen]

Henning Meier-Geinitz <henning@...> writes:

>> Having now read a previous mail about the 4490 I'm guessing that
>> the same applies to the 3490, ie no go..are my fears justified?
>
> That scanner isn't even in our list. Can anyone provide more details
> (vendor/product ids, chipset)? Support is planned for the epkowa
> backend, see the list archive of this list for details.

The 4490 is in the same category as the 3170 and 4180.
USB vendor/product ID info: 0x04b8/0x0119.

Hope this helps,
--
Olaf Meeuwissen                          EPSON AVASYS Corporation, LAN
FSF Associate Member #1962           sign up at http://member.fsf.org/
GnuPG key: 6BE37D90/AB6B 0D1F 99E7 1BF5 EB97  976A 16C7 F27D 6BE3 7D90
Penguin's lib!       -- I hack, therefore I am --               LPIC-2

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Olaf Meeuwissen]

Olaf Meeuwissen <olaf.meeuwissen@...> writes:

> Jim MacLeod <j75trym@...> writes:
>
>> I'm new to scanning and relatively new to Linux. Searched the web
>> for info and got the impression, which others confirmed, that
>> generally anything epson works with Linux, but maybe not the 3490!
>> sane-find-scanner gets vendor and product code but scanimage -L
>> finds nothing.
>>
>> Having now read a previous mail about the 4490 I'm guessing that the
>> same applies to the 3490, ie no go..are my fears justified?
>
> The same does not apply.  The 4490 and 3490 are very different
> scanners.  The 3490 is partially supported by the latest snapscan
> backend.  Also, iscan will support the 3490 with the next version
> (1.17.0) using a closed source plugin.

Sorry, forgot to mention the URL.

  http://www.avasys.jp/english/linux_e/

BTW, iscan-1.16.0 saw the light of day yesterday.  It adds support for
a small business MFP marketed in Japan.  That's probably not so very
interesting for most of you but what may be more interesting is that
there are now also experimental gcc-3.4 or later packages.

For those who missed 1.15.0, that version added support for the Stylus
Photo RX700 and the AcuLaser CX11.  Moreover, iscan uses GTK+ 2.x by
default as of that version and can be used as a GIMP 2.x plugin.

>> Vuescan which does support the 3490 but I've got to learn more about
>> usb first before going down that road - or is there hope.
>
> Hope this helps,
--
Olaf Meeuwissen                          EPSON AVASYS Corporation, LAN
FSF Associate Member #1962           sign up at http://member.fsf.org/
GnuPG key: 6BE37D90/AB6B 0D1F 99E7 1BF5 EB97  976A 16C7 F27D 6BE3 7D90
Penguin's lib!       -- I hack, therefore I am --               LPIC-2

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Oliver Schwartz writes:

> Hi,
>> I don't have access to that information right now, so I'll check
>> that tonight also.  It is whatever firmware came with the scanner -
>> esfw52.bin.

> yes, but at least for other epson scanners these firmware files are
> available in different versions. The filename is always the same.

$ strings esfw52.bin  | tail
Thu Apr 14 2005 14:32
1.08 u
A0~      
 4{
r {o0r
~rSP
u89u9
u89u9
EPSON   GT-F520         1.08
qThu Apr 14 2005 14:32


> Another thing that I'm interested in is whether only the size of the
> scan area is wrong or if the origins of the scan area are also not
> set correctly (i.e. options -l and -t in scanimage)

-x and -y are reduced by about 25% - i.e. I ask for 100mm and get only
75mm.

-l and -t are ditto reduced by about 25% - getting only 75mm when asking
for 100mm.

> If it is always the same ratio of real distance to scanned distance
> (also for -l and -t) it is most likely that the calculation of the
> scan area is not handled correctly. That could be corrected by
> changing the value of pos_factor in function set_window() in
> snapscan-scsi.c.

That would seem to be the case here.

> For the guy that developed the patch that didn't help, though.

Hm.  Is that the guy that you say below has the scanner working for him,
or is it someone else?

>> My main interest at this point is to understand whether this issue
>> is something that can be solved with some patch to the backend, or
>> whether it's inherently related to the scanner's firmware or
>> something that is Epson-proprietary that the SANE developers won't
>> get access to. Is anyone able to shed some light on this issue?  

> I don't think that this is an unsolvable problem. It may be necessary
> to do some deeper analysis and maybe get some USB traces from the
> windows driver, though.

How do I go about getting such traces for you (or whoever it is who'll
likely be working on this)?

> Also keep in mind that the scanner seems to work for at least one guy,
> so the solution can't be that difficult :-)

I certainly hope that is the case!  What did that guy do to make it work?
I.e. scanning over the whole area?

(On a side note: Is brightness supposed to work for lineart mode?  As it
is, it tends to make Lineart mode scans too dark - things that are
clearly (but somewhat faded) white paper gets smothered in black.  My
UMAX Astra 1200 didn't have that problem at all.)

Thank you,

Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Levente Novák writes:

> On Thu, 2005-09-01 at 00:54 -0700, Bjorn Solberg wrote:
>> > Another thing that I'm interested in is whether only the size of the
>> > scan area is wrong or if the origins of the scan area are also not
>> > set correctly (i.e. options -l and -t in scanimage)
>>
>> -x and -y are reduced by about 25% - i.e. I ask for 100mm and get only
>> 75mm.
>>
>> -l and -t are ditto reduced by about 25% - getting only 75mm when asking
>> for 100mm.
>>
>> > If it is always the same ratio of real distance to scanned distance
>> > (also for -l and -t) it is most likely that the calculation of the
>> > scan area is not handled correctly. That could be corrected by
>> > changing the value of pos_factor in function set_window() in
>> > snapscan-scsi.c.

> AFAIK the Epson is a 3200 dpi scanner while e.g. 2480 is only 2400 dpi.
> 2400/3200=0.75.

> Couldn't be this the problem?

Duh.

So maybe it's just a matter of finding the part of the code that governs
this and set up a separate configuration for the 3490.

Yet another issue is that scanimage seems to hang if I try to scan with
resolutions over 1200 - could that be related?

Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Oliver Schwartz]

Hi,

> > If it is always the same ratio of real distance to scanned
> > distance (also for -l and -t) it is most likely that the
> > calculation of the scan area is not handled correctly. That could
> > be corrected by changing the value of pos_factor in function
> > set_window() in snapscan-scsi.c.
>
> That would seem to be the case here.

Yes, so maybe it's enough to just change that value for the 3490. It's
set in line snapscan-scsi.c, line 810. The value is currently set to
1600 for resolutions below 1600, 3200 for resolutions above. If your
distances are off by 25%, try setting it to 1600/0.75 = 2130.

> > For the guy that developed the patch that didn't help, though.
>
> Hm.  Is that the guy that you say below has the scanner working for
> him, or is it someone else?

Yes. Unfortunately I've lost contact to him, i.e. he doesn't answer my
mail.

> > I don't think that this is an unsolvable problem. It may be
> > necessary to do some deeper analysis and maybe get some USB
> > traces from the windows driver, though.
>
> How do I go about getting such traces for you (or whoever it is
> who'll likely be working on this)?

You can use usbsnoop:  http://benoit.papillault.free.fr/usbsnoop/
The files tend to get very large, so chose a very small scan area and
a low resolution. You can send them to me in private mail.

> > Also keep in mind that the scanner seems to work for at least one
> > guy, so the solution can't be that difficult :-)
>
> I certainly hope that is the case!  What did that guy do to make it
> work? I.e. scanning over the whole area?

He developed the patch. His last mail stated it works, so I applied it
to CVS.

> (On a side note: Is brightness supposed to work for lineart mode?
> As it is, it tends to make Lineart mode scans too dark - things
> that are clearly (but somewhat faded) white paper gets smothered in
> black.  My UMAX Astra 1200 didn't have that problem at all.)

Yes. The parameter is called "Threshold" in lineart mode. It's
available in the Standard Options window in xsane or as option
--threshold in scanimage.

/Oliver


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Oliver Schwartz writes:

> Hi,
>> > If it is always the same ratio of real distance to scanned
>> > distance (also for -l and -t) it is most likely that the
>> > calculation of the scan area is not handled correctly. That could
>> > be corrected by changing the value of pos_factor in function
>> > set_window() in snapscan-scsi.c.
>>
>> That would seem to be the case here.

> Yes, so maybe it's enough to just change that value for the 3490. It's
> set in line snapscan-scsi.c, line 810. The value is currently set to
> 1600 for resolutions below 1600, 3200 for resolutions above. If your
> distances are off by 25%, try setting it to 1600/0.75 = 2130.

The CVS snapshot I downloaded for 20050815 didn't contain that patch.
And the CVS version doesn't work for me now (whether I make that
modification or not), I alternately get messages like

scanimage: open of device snapscan:libusb:001:003 failed: Error during device I/O

and

scanimage: no SANE devices found

But even so, scanimage -L reports

device `snapscan:libusb:001:003' is a EPSON EPSON Scanner1 flatbed scanner


Reinstalling the version from 20050815 (rev. 1.32 of snapscan-scsi.c)
makes the scanner work again though, but with the same scaling (and
brightness) problem as before of course.


[...]

>> > I don't think that this is an unsolvable problem. It may be
>> > necessary to do some deeper analysis and maybe get some USB
>> > traces from the windows driver, though.
>>
>> How do I go about getting such traces for you (or whoever it is
>> who'll likely be working on this)?

> You can use usbsnoop:  http://benoit.papillault.free.fr/usbsnoop/
> The files tend to get very large, so chose a very small scan area and
> a low resolution. You can send them to me in private mail.

I'll get you that shortly.

[...]

>> (On a side note: Is brightness supposed to work for lineart mode?
>> As it is, it tends to make Lineart mode scans too dark - things
>> that are clearly (but somewhat faded) white paper gets smothered in
>> black.  My UMAX Astra 1200 didn't have that problem at all.)

> Yes. The parameter is called "Threshold" in lineart mode. It's
> available in the Standard Options window in xsane or as option
> --threshold in scanimage.

I get (with the 20050815 version)

scanimage: attempted to set inactive option threshold


Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Oliver Schwartz]

Hi,

> The CVS snapshot I downloaded for 20050815 didn't contain that
> patch. And the CVS version doesn't work for me now (whether I make
> that modification or not), I alternately get messages like
>
> scanimage: open of device snapscan:libusb:001:003 failed: Error
> during device I/O
>
> and
>
> scanimage: no SANE devices found

Can you send me a debug log? Run
export SANE_DEBUG_SNAPSCAN=255
then
scanimage 1>/dev/null 2>debug.log

> > Yes. The parameter is called "Threshold" in lineart mode. It's
> > available in the Standard Options window in xsane or as option
> > --threshold in scanimage.
>
> I get (with the 20050815 version)
>
> scanimage: attempted to set inactive option threshold

It's supposed to be active in lineart mode only. Running
scanimage --mode=Lineart --threshold=70
works fine here (although on another scanner model).

/Oliver


--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...

[Bjorn Solberg]

Oliver Schwartz writes:

> Hi,
>> The CVS snapshot I downloaded for 20050815 didn't contain that
>> patch. And the CVS version doesn't work for me now (whether I make
>> that modification or not), I alternately get messages like
>>
>> scanimage: open of device snapscan:libusb:001:003 failed: Error
>> during device I/O
>>
>> and
>>
>> scanimage: no SANE devices found

> Can you send me a debug log? Run
> export SANE_DEBUG_SNAPSCAN=255
> then
> scanimage 1>/dev/null 2>debug.log

(Sent in private email.)

>> > Yes. The parameter is called "Threshold" in lineart mode. It's
>> > available in the Standard Options window in xsane or as option
>> > --threshold in scanimage.
>>
>> I get (with the 20050815 version)
>>
>> scanimage: attempted to set inactive option threshold

> It's supposed to be active in lineart mode only. Running
> scanimage --mode=Lineart --threshold=70
> works fine here (although on another scanner model).

Ah, I had the --threshold argument before the --mode argument - I didn't
realize there was a required order and didn't test any permutations.
Thank you for clearing that up for me.

Bjorn.

--
sane-devel mailing list: sane-devel@...
http://lists.alioth.debian.org/mailman/listinfo/sane-devel
Unsubscribe: Send mail with subject "unsubscribe your_password"
             to sane-devel-request@...