[sane-devel] New backend for a not yet exisiting device

[Torsten Wagner]

Dear Dev-Team,

first of all I'm a complete newbie in terms of scanners resp. the development 
of sane. I hope you can help me a bit to sort out some questions.

Background: We are going to develop a new kind of scientific measurement 
device which acts similar to a scanner.  Our measurements are latterly 
resolved e.g. we can create a pixel by pixel map whereas the measurement 
signal might be represented by a colour. This is, so fare well known already 
for similar devices already (Actually, it is not different to a normal 
scanner). Our reading unit can read line by line and thus we are thinking to 
use the mechanical set-up of a normal scanner to acquire a complete picture.

Idea: For sure, it would be perfect, if we can address this type of new 
scanner in the same way like a normal scanner. E.g. by TWAIN-Protocol or by 
SANE. This guarantees to capture measurements many kind of different 
software. 

Problem: I have totally no idea how the communication between a scanner and 
the computer works. I read about intelligent scanners which just get commands 
like "scan with 600dpi one A4 page" and about dumb scanners which requires 
commands like "turn on stepper motor for next line, scan line, etc." 

Is there any information available which might help me to identify which basic 
scanner set-up might be the best for our purpose. I guess the dumb version 
gives us access to lower level functions which we might need ?Any kind of 
this hacks known already ?

Does somebody know a scanner brand or a specific model which is easy hackable 
for our purpose. Is there one out with full open/known programming interface 
or does this require always a reverse engineering of all the different 
commands ? Is there a sane-dev favourite scanner due to the easiness of use 
and addressing ? A USB-model might be preferable but a LPT-based model might 
be easier to access ?!

I appreciate any kind of information and help.


Best regards

Totti