[sane-devel] I have the Epson Perfection 2480 and 2580 for testing

Till Kamppeter till.kamppeter@gmx.net
Fri, 22 Oct 2004 20:41:29 +0200

Here are some more debug logs, all for the 2580 (lower part with film 
feeder lid):


To get these logs the scanner was connected and turned on and then the 
following command lines executed:

[root@majax g]# SANE_DEBUG_SNAPSCAN=255 scanimage > 1.pnm 
[root@majax g]# ll 1.pnm
-rw-r--r--  1 root root 0 Oct 22 13:50 1.pnm
[root@majax g]# lsusb
Bus 001 Device 020: ID 04b8:0121 Seiko Epson Corp.
Bus 001 Device 003: ID 0451:2077 Texas Instruments, Inc. TUSB2077 Hub
Bus 001 Device 002: ID 04f9:0016 Brother Industries, Ltd
Bus 001 Device 001: ID 0000:0000
[root@majax g]# SANE_DEBUG_SNAPSCAN=255 scanimage > 1.pnm 
[root@majax g]# ll 1.pnm
-rw-r--r--  1 root root 26849629 Oct 22 13:58 1.pnm
[root@majax g]# SANE_DEBUG_SNAPSCAN=255 xscanimage >eps2580debug5.log 2>&1

The scanimage commands are non-interactive and only on the second the 
scan succeeded (file contains the correct image). The first did not 
succeed because there the firmware was uploaded and the scanner 
reconnected with a new address as told already in the previous posting. 
The third command was an interactive xscanimage session. The following 
steps were done:

1. Normal preview, failed as described in the previous posting: fast 
sweep of scan head, no correct image.

2. Scan of a very small part of the image area. Succeeded, file contains 
expected image.

3. Preview with TPU, took VERY long until it started, head sweep sounded 
correct, but color components seem to be misaligned.

4. Scan with TPU, took long but seems correct.

5. Back to normal flatbed mode: Preview scan still broken.

6. Normal scan of full area at 100 dpi absolutely correct.]


Till Kamppeter wrote:
> Oliver Schwartz wrote:
>> Hi,
>>> Why does the frontend want to set the gray gamma table to (nil)?
>>> [snapscan] sane_snapscan_control_option (0x823a6b0, 24, 1, (nil),
>>> 0xbfffdffc)
>>> Maybe I don't understand something here but that looks strange.
>> This seems to cause the crash (which is not that surprising). For some 
>> reason the frontend passes an invalid pointer.
> I have tried now with xscanimage and have interesting results:
> The 2480 and 2580 consist of a lower part with the glass and scanning 
> head and a lid with the transparency unit. The lower parts of both are 
> absolutely identical, including the USB vendor/product IDs. The 
> difference between the 2480 and 2580 is made up by the lids. The lid of 
> the 2480 has a simple light which is turned on in transparency mode. You 
> insert the film or slides in a special frame and put this on the glass, 
> so film or slide are placed under the light and parallel to the 
> scanner's borders. The 2580 has an automatic film feeder. You remove the 
> white background from the inner side of the lid, close the lid and open 
> a little door on the upper side of the lid. There you insert a piece of 
> 35-mm film and the scanner pulls it in automatically. For slides there 
> is a frame for the scanner glass again, but a different one.
> In principle the scanners work with SANE and the "snapscan" driver, but 
> there are some problems:
> - You need the scanner's firmware, as for all scanners used with the 
> "snapscan" driver.
> - There are long delays between starting the scanner software and the 
> GUI opening or the scan beginning. Messages like
> [snapscan] Scanner warming up - waiting 33 seconds
> appear on the console from where the scanning frontend was started.
> - The biggest problem is that the USB device address (like 
> libusb:001:015) changes when the firmware is uploaded, it seems that 
> after the upload the scanner is disconnected and then reconnected and so 
> it gets a new ID.
> This means that the very first start of a frontend (after 
> connecting/turning on the scanner) lets the GUI open, but trying to 
> preview of to scan gives "Invalid argument". Comparing the output of 
> "lsusb" reveals that the USB device address changed at the moment where 
> the firmware upload finished. On subsequent runs of scanning frontends 
> the firmware is not uploaded again and so the address does not change. 
> So one can scan when one closes the frontend and opens it again. Perhaps 
> one should do modifications on the "snapscan" backend to take the 
> reconnection/address change into account.
> - The lid of the 2580 is not compatible with SANE. If the lid is in use 
> with any of the two lower parts (2480 and 2580) the preview does not 
> work. The head moves shortly and the preview is a stripy dirty white. 
> Normal scanning does work at least in flatbed mode. Switching to the 
> transparency unit makes the frontend freezing.
> Both lower parts work without problems when the lid of the 2480 or no 
> lid at all is connected. With the lid of the 2480 one can switch to the 
> transparency unit and scan transparent documets correctly. Without lid 
> the button for the transparency unit simply does not appear.
>    Till
>> Can you try another frontend (e.g. xscanimage)? It may also help to 
>> delete any xsane preferences (stored in ~/.sane/xsane IIRC).
>> /Oliver