Slashdot Mirror
Scanner Server?
chuckw asks: "Does anyone know if there are any open source projects that are working on building a scanner server and client? I have a pretty nice document scanner plugged into my Linux server and I want to give everyone in my house access to it, much like a regular document printer. Network bandwidth isn't really an issue (100baseT). I personally would like to start such a project but I don't want to duplicate anyone's effort. I've looked all over and can't find any currently running projects doing this. If this isn't currently being done, do you want to join the project? Do you see any major technical issues preventing such a project? "
I don't see how having the scanner shared can be any more efficient than having it local to one computer (your scanning workstation) and just having it save files to a directory that is shared via samba or nfs or whatever. With a shared directory, you also have the advantage of having *several* computers get at the document from one centralized location without having to re-scan it each time.
Compare this to printing: what would you rather have (in an ideal situation), a printer queue on each workstation, or one centralized one on the computer that is connected to the printer?
It's only software!
Of course, the way I scan usually involves having to reset the page since it is crooked, cropped, etc., so I would end up running back and forth between the remote computer and scanner too much.
Also, if the "server" is a computer somebody else is using, you wouldn't have to kick them off everytime you wanted to scan something.
And its late at night here, so maybe I'm missing the point, but isn't the server-client scanning setup closer to the single queue on the centralized computer example than the queue on each workstation, or is that what you were trying to say? If it is, then then how is a seperate queue per workstation better/possible? If none of that made sense, don't worry it's probably me, I think I'm going to bed now.
Printers have pristine digital data; The person who sends it to the printer KNOWS what the printer is goint to output. When you scan something from an analog source, the data isn't pristine and ALWAYS needs adjustment. I have done a fair bit of commercial image processing (legal documents). Stick fifty documents into the ADF, half illegible faxes, greenbar 8 1/2 x 11, promotional flyers printed on bright red stock, a couple of big glossies and the source code to DeCSS. Scan all of them at the same set of threshold, contrast stretch and balance. I bet dollars to donuts only the source to DeCSS comes out 100%. When you are sitting at the scanner, or have 'local' access to the scanner, you can activly tweak the image stream on a per document basis, preview and scan, and if needed, (and it is available) send it back through the ADF loop for a second pass. But you can always just tweak at the scanner, right? Nope. I have seen 'standard' installations that required three imaging techs on NT boxen to scan the images just for export to the Irix workstations for processing. See, Irix didn't like the scanners that were required (damn you, Fujitsu!) and no other platform was capable of running the custom applications needed. It was an endless tirade of 'I need more X in that last one' or 'can you pull that sucker out of pseudo-zoom, please?? The noise is killing me!'
.sig: Now legally binding!
Simple, you have your lackey go put in the document. Seriously most network scanners have a Automatic Document Feeder, so you put in a stack of pages go back hit scan let it scan all 100 pages or so. Also it is cheaper to buy 1 really good high end scanner with a ADF and network it than to buy 20 cheap scanners.
Where are we going and why are we in a handbasket?
Well, that depends. If your scanning a million different documents without an ADV, it won't work. Others have pointed out the problems. (contrast, diagnal...)
But be creative. My scanner is scsi, and my main machine doesn't have a scsi adaptor. (Accually it does, but it is wide scsi, and scanners ahve a bad repuatation for tieing up the scsi bus which isn't a good idea on a multitasking OS) I put the scanner on the old 386 (which seems to keep getting new functions) and then used network scanning to work.
At one place I lived we had several computers scattered throughout the house. We didn't need many scanners though, network scanning [would have] allowed us to preview scan on the slow machine with the scanner (a macII), and then use a faster machine once we had things straight. As it was we continually ran out of memory and harddrive space on a machine that spent more time processing the data then scanning. If I had done the final scans upstairs it would have been faster overall.
Perhaps I've missed something... but I don't know what. I could see if it was one scanner in the middle of a bunch of computers -- perhaps in a computer lab -- but over a whole house?
Please enlighten me.
--
First, my disclaimer: I have never tried this with Linux. All my experience with this has been with the OS/2 port of SANE, and eventually I had to give up because I couldn't make it work with my Microtek E3 scanner. That said, I believe that SANE ought to do what you want. It comes with a program called SANED - run that on the machine with the scanner and then, provided you use SANE on the clients to do the scanning it should all work beautifully, as if the scanner was locally attached. SANE is a command-line tool, but there are, I guess, plenty of X front-ends for it etc. HTH