Hi, I am still playing with Acer ScanWit 2720S film scanner and I found out some protocol details that allow me to scan. I am about to start writing a quite restricted SANE backend for this thing. Though, there are some things that are still quite unclear to me. I wonder if some of you have any ideas of what could be happening. So, I list these problems here.
1. Scan speed/mode. When I issue a SET WINDOW command and set the resolution to 2700 (maximal optical resolution of the scanner), it scans at speed comparable to what I get with existing drivers (windows and VueScan). Then I reduce the resolution to 1350, and the scan is almost as slow as before. Why? 2. (somewhat related to 1) When I reduce the resoultion even more -- to 675 -- the scan *still* takes almost as much time as the scan at 2700. Moreover, the resulting file, while having proper dimensions, contains only part of the image; plus, it seems to have been stretched in vertical direction (every row doubled or like that). Did anyone experience similar effects on some other scanner? 3. I traced VueScan's SCSI dialog, and figured out just about every command it issues, except for one part. Maybe it is something common for scanners, I just have little experience and/or knowledge. The protocol is like this: 1. RESERVE 2. Set focus 3. SET WINDOW 4. Upload gamma tables 5. INQUIRY 6. [strange part] 7. SCAN 8. Set frame holder to desired frame 9. READ.... 10. RELEASE The ``strange part'' 6 looks like this: READ with DT=0x80 and DTQ=0, 61056 bytes READ with DT=0x80 and DTQ=0, 0 bytes READ with DT=0x82 and DTQ=0x10, 7650 bytes (repeats 15 times) SEND with DT=0x82 and DTQ=0x01, 7650 bytes The data sent and received had little meaning to me -- it contained a block of bytes near to 0xff in one place and a block of bytes near to 0x3c in another. What could this be? This could not be the autofocusing sequence because I already know how to do autofocus. These are the main problems I have. I also have one question: where should I start writing the new backend? It seems a good idea to take some existing one as a template -- but I don't know what... Thanks for reading all this, --Max Ushakov.
