On Tue, Jul 9, 2013 at 3:34 AM, Hakan Bay?nd?r <hakan.bayindir at pardus.org.tr> wrote: > Hi Allan and all, > > Thanks for the information. I've scanned ~100 pages with the scanner using > XSane and scanimage with various settings, I'm sharing my findings below. > I'll try to keep the scanner as much as I can to aid its development and > testing, so I'd be glad if you can guide me.
Thanks for this effort. > > - Default page size of the printer is set to letter. For A4 it's wide and > short. A default page size with letter width and A4 length would be better > IMHO. That would cause the length-based double feed detection to not work correctly with letter scans. In all the ADF backends I have written, you must set page-width and page-height before you set the scan dimensions. I suppose what we need is a config file setting which will set the default paper size for the backend, so that it can be localized. I will think about that. > - XSane is making cropping errors with media smaller than A4. I've scanned a > set of A5 pages. Before, I got a preview from Xsane and set the margins. > Right side of the pages are cropped and mirrored to left side. Mirrored part > was approximately 1cm. Gap between mirrored part and actual left margin was > also ~1cm. This sounds like a bug- please attempt to reproduce this with scanimage, and send me the command you use, and the image. > - Length based double feed detection works in XSane. Thickness based didn't > work, but I'm not sure if XSane enables it or not. The scanner may not have a thickness sensor- there is no way to detect that, IIRC. > - I had to give both X & Y and page-width/length in Scanimage to get an > accurate scan. I don't know it's expected, but the resulting images were > literally perfect. That is expected. These options should also be used in XSANE. > - Batch mode of the scanimage works as expected. XSane doesn't do blind > batch if I understood correctly. XSANE has a page counter, which you can increase to a large number. But overall, XSANE is not optimized for document scanning. gscan2pdf works better. > - Color, gray, lineart scanning works as expected. 600DPI color scans are > razor-sharp. Greys are well defined and very sharp. Line art is very good as > well. > - Scanner seems to can work at its top speed incl. duplex (tested in 200DPI > Gray scans, pnm output). This sounds good. We are making the lineart in software. > - I didn't tried to work with the buttons on the scanner. That would require an external program to monitor the buttons. Something like scanbd. > - I'm not an expert on sheetfed scanners, but requiring to put document > title-down (rotated 180 degrees) felt awkward a bit. If it's expected, it's > OK. If not, driver needs some page flipping code. That is normal. Adding rotation code would slow scanning, because we would have to buffer the entire image. > - I was not brave enough to test staple detection, sorry. It may not work on this scanner anyway :) > These are all of my findings for now, I can run additional testing if you > want. Last but not least, I'm a developer so I can try new things if you > need. Let's see if we can get the cropping issue you mentioned above worked out. allan -- "The truth is an offense, but not a sin"
