CanoScan 9000F support in SANE

[dpic]

Great to see work being done on this! What's the progress so far? Is there a bug report i can subscribe to? Someone files this on launchpad:
https://bugs.launchpad.net/ubuntu/+sour ... bug/601652

[shugyo]

Hello dpic,
If you want the latest patch, contact me by PM. I am happy to announce that after the last 3 days of testing, one of the wonderful test gentlemen, Eddy De Greef, sent a corrected code patch for the 9600dpi mode (TPU-only, flatbed does not have 9600dpi mode) and the scanner now works for all modes.

There is a problem with the maximum buffer size that the driver can handle, so large color scans at 4800 or 9600 dpi may only result in a portion of the image being delivered. We are working on isolating this problem.

The issue of getting the code to support this scanner into SANE is one we need to tackle after this, as there are quite a few big changes that need to be vetted, and perhaps alternatives to the current implementations written and tested if they do not meet the requirements of the SANE development team. That is why in the meantime I am happy to send the patched code (which also contains many other patches not yet in SANE, such as the 4800dpi mode for the MP960, etc.)

Best regards,
Gernot Hassenpflug

[shugyo]

Dear all,

I am delighted to notify the readers of this thread that the 9600dpi mode for the CS9000F works perfectly now, as do other modes.

Also, the problem with limited image size (2^32 byte limit) has been solved. The changes should go into SANE CVS in the next few months, but as there are many many changes, users wishing to make use of the CS9000F, MP960, MP810 and others with 4800 or 9600dpi resolution modes, please contact me directly for patched code files.

Best regards,
Gernot Hassenpflug

[shugyo]

This is a short update on development, basically sorting out minor problems that get reported.

1) latest code (not yet sent out) is independent of whether 32-bit or 64-bit architecture is used.
2) cleaned up and added many comments (thanks to all the gentlemen who added these, and developed the code).
3) one minor issue on TPU area y-offset is being sorted out.

There are now a total of 20 other persons who have received the code, and many of them have either sent in code changes, bug reports, or tested functionality in depth. I am deeply indebted to them all, as I myself do not have the scanner and would have been unable to develop so quickly as has transpired. The progress in the past 2 weeks has been phenomenal.

I have already begun the process of applying for the code to be included in SANE CVS, beginning with the infrastructure changes in the pixma backend generally, before moving on to the diff of the sub-driver (which includes also MP960 and MP810 support).

Best regards,
Gernot Hassenpflug

[shugyo]

Hi all,
Short update: all minor bugs seem to be sorted out in the code for the CS9000F (and MP810, and MP960).

I am distributing this code myself for now, so please PM me if you want it.

I also started the process of submitting diffs to the maintainer for inclusion of the code into CVS. After massaging the code to not clash with the CVS code too much, the diff is 740 lines long.

Hopefully the next 1-2 weeks will see it reviewed & included in CVS.

[shugyo]

Hi all,
An update on the SANE CVS inclusion.
I will create a sub-driver which will include the following devices:
MP810
MP960
Canoscan 8800F
Canoscan 9000F
and any others that I find in the intervening period requiring similar programming (I have an MP610 to test right now).
When the sub-driver is completed I will submit the files to the maintainer for review. I expect to complete this by 21 January.

In the meanwhile, I will continue to distribute the combined driver that I currently have.

Best regards,
Gernot Hassenpflug

[shugyo]

Dear all,
I have some more news, some good, some not so good.

First, the bad news: the Canoscan 9000F will NOT be in the next SANE minor release 1.0.22, for which the freeze deadline has already passed. This means that the code for supporting this scanner will be distributed by myself until such time as it is in CVS (see below), after which it will be available in the SANE CVS version.

The simple reason for non-inclusion is that the existing sub-driver code is too complex, and a new sub-driver should be written (extracted from the current sub-driver in which the code is now located) which can be maintained independently from all the other scanners in the existing sub-driver. For those who are unfamiliar with the pixma backend, there are several sub-drivers, and the pixma_mp150.c one to which we added the CS9000F support also needs to support about two dozen other scanners from at least 4 different generations of Canon's pixma protocol.

And the reason I could not yet write a sub-driver is a combination of time constraints, and the recent discovery of still one remaining issue in TPU mode, regarding the preview scan.

Second, the good news. the above problem in TPU mode was fixed today, and after some tests this week I can get down to writing a proper sub-driver for the CS9000F, which will also support the CS8800F, the MP810, and the MP960. These devices will then be removed from the existing sub-driver. I will submit my code to the maintainer to approval hopefully this weekend, and report in this thread on developments in regard to adding the code to CVS (probably only possible shortly after the release of SANE 1.0.22 in mid-February).

If you are interested in the code, or want to help with testing & development in any way, please contact me. Also, if you have any scanners with TPU mode (that is, CCD scanners that can handle negative and positive film scanning) please feel free to contact me, as I am keen to find out whether there are other devices with TPU capability that need the code written to support the CS8800F, CS9000F, MP810 and MP960.

Best regards,
Gernot Hassenpflug

Edit: cleaned up thread to reflect current status

[shugyo]

Short update:

A bug with TPU preview mode offset fixed over the weekend. Also, SANE 1.0.22 is ready to be released, so the git repository for 1.0.23 should soon be ready. Then we can start the process of sending diffs for the new sub-driver incorporating the support for the 9000F, 8800F, MP960 and MP810.

Best regards,
Gernot Hassenpflug

[ziemlich]

I just found this thread about the great effort has been expended to make the 9000F in SANE usable. It makes me happy, because I've been hunting the 8800F since a month because of its Complete support, but it's no more available in Europe. So, I'm about to buy the 9000F. I would ask about the driver, does it support all features which are available in W7? For example I want to scan my long-ago made negative films. Does the scanner uses the FARE correction in TPU mode automatically?

[samfreed]

Just got one of these scanners, and can't find the stuff in GIT. HELP!

[saedelaere]

How is progress going for the Cannon 9000F, is it possible to get the prerelease driver from you? How about scanning negatives, is the transparency unit working?

Best regards and many thanks for your efforts

[shugyo]

I just found this thread about the great effort has been expended to make the 9000F in SANE usable. It makes me happy, because I've been hunting the 8800F since a month because of its Complete support, but it's no more available in Europe. So, I'm about to buy the 9000F. I would ask about the driver, does it support all features which are available in W7? For example I want to scan my long-ago made negative films. Does the scanner uses the FARE correction in TPU mode automatically?

Hello,

It is unfortunate for us all that so far no good way to implement FARE has been found---so this is not available. Some fellow enthusiasts were working on FARE, as was I for a short while, but time limitations prevented a good resolution, although some success was achieved. The problem of registration of the IR and color images was one issue (it is not constant, but needs to be adjusted depending on the part of the image); also, the resolution of the IR image was only a subset of the resolutions available in color, so adjustment of the IR image was also needed (interpolation). Finally, some results were obtained, but the code is nowhere near production ready and consists of various test programs which are not related to the current SANE code.

Regards,
Gernot Hassenpflug

[shugyo]

How is progress going for the Cannon 9000F, is it possible to get the prerelease driver from you? How about scanning negatives, is the transparency unit working?

Sure, the TPU works fine---the results are much as for the 8800F.

Calibration is not done properly yet, so that is something that could definitely use work (same for 8800F).

Anyone that wants the code is welcome to write to me (PM me).
I hope to find time in July to submit a modified driver to SANE developers for inclusion in CVS.
Regards,
Gernot Hassenpflug

[All]

Hi Gernot,

Great work you guys are doing here. I'm amazed at the effects of opensource, I'm not so young any more, and fascinated by you guys working so passionately to help one another (but that's another story).

I want to buy the CanoScan 9000F. I have openSUSE 11.4 64Bit installed with Xsane and sane 1.0.22 . Is your driver already included? If not, I will gladly test with openSUSE for you if you can direct me. I will send you a PM for the driver.

Regards,

All

[nananabatman]

Thank you for your work! Looking forward to finally using the 9000F on linux!

[Rodboudreaux]

I was on the team that developed the PIXMA MP150 backend and have it running with the latest xsane as root.
my job was testing. I have a CanoScan 9000F. My skills are limited but I follow direction well. I run under Ubuntu.
How can I help.

Rod

[shugyo]

I've started to split the driver (pixma_mp150) into two: original pixma_mp150 without support for the 9000F, and a new pixma_mp810 with the 9000F support.

Probably the splitting, cleaning up of code, and verification before it is put into the git repository of SANE will take some time (2 weeks possibly, although I have no idea really), but the process has begun. So hopefully early next year you'll all be able to get the driver support for the 9000F via the development version of SANE, and the support will be included then in the next SANE release.

[shugyo]

Good news: in the night of 18-19 December 2011 I submitted the new sub-driver, associated patches to other SANE files, and documentation patches, to the SANE mailing list. The process of incorporating the new programs will take some time (presumably code review, then corrections, then finally commitment to the git repository), but the process is now safely in motion.

[nananabatman]

great news!

[shugyo]

It has been a whilte, so just to complete the thread (sorry, could not get numbered lists to work):

• (1) the code is in the SANE git repository, so anyone that has the scanner can be confident that compiling SANE from git will give the required support---also note that the development is ongoing (see point 2 below).
• (2) there is a code maintainer, who is very competent and actually has the scanner (unlike myself, who only has the previous 8800 model), and is interested in improving the code incrementally.
• (3) please use the SANE mailing list (sane-devel@lists.alioth.debian.org, see http://www.sane-project.org/mailing-lists.html for registration information) to contact the developers, this is by far the best place to get support and have questions answered.
• (4) I cannot currently give any support for SANE, as I am totally underwater maintaining and developing the Canon backend for the gutenprint project (inkjet driver support).
• (5) I do not read this forum or thread regularly.
• 6) if you wish to ask me (Gernot Hassenpflug) for help in private messages, I will certainly do my best to give you advice and tips, but unfortunately recently I had experience being messaged by people with absolutely no respect for my time, so with all due respect to new-comers: please read the above points first.
• (7) If you message me please be prepared to do some work on your own (I will gladly provide links and advice); if you message me for further support please demonstrate that you have done relevant work yourself, and also why the normal channels are not sufficient. I appreciate that people have diferent capabilities and expectations, but also that it is a privilege to obtain help and support from others who freely give their time.

Many thanks,
Gernot Hassenpflug