[sane-devel] gain/offset tables

[nathan rutman]

stauff at guarana.org said:
> The FB330P/FB630P behave the same way, and probably have the same CCD
> sensor.  If you can understand Simon's code, then feel free to use it.
>  The adjust_output() method in canon_pp-dev.c is probably most useful
> for your purposes.  It takes an RGB-ified image block and applies the
> calibration data to bring the colours to the correct levels.

Yeah, I did look at it - but it's not as complicated as I want. :)

The FB630u's LM9830 has the ability to correct for this stuff
automatically via on-board offset/gain and gamma tables. It also can
adjust the analog gain for R,G,B channels which could theoretically be
used to get a better full-color range.

But so far, it just looks like everyone (FBx30P, Microtek, Primax) does a
post-scan software-based linear correction.  Does anyone know what other 
scanners use the LM9830?  Anyone else use these specific features?



Original message:
> Thinking about the calibration routine for my scanner (Canon FB630U),
> seems to me this is probably a common issue.
>
> There's a black and a white scannable area inside the housing of my
> scanner.  It looks like I need to convert scans of these areas under 3
> different lighting conditions into the per-ccd-element gain/offset
> tables.  ("Looks like", because that's what the Windows driver seems to
> do.)
>
> Anyone's scanner do something similar?  Got code or math I can steal?



--
nathan at gordian.com