[sane-devel] Reflecta DigiDia 5000 not found

Reinhard Kotucha reinhard.kotucha at web.de
Fri Jan 5 21:53:31 UTC 2018


On 2017-12-29 at 10:40:46 +0900, Olaf Meeuwissen wrote:

 > Reinhard Kotucha writes:
 > 
 > > [...]
 > > I could probably improve the results if anybody tells me what these
 > > values mean and how to apply them:
 > >
 > >   FilmMatR    1.4234   -0.3230    0.0061   -29.7088;
 > >   FilmMatG    0.0600    0.9677    0.0719   -31.3147;
 > >   FilmMatB    0.1562   -0.5104    1.4530   -28.9059;
 > 
 > I'm not sure about the rightmost column but this is a color correction
 > matrix.  You apply it on the incoming (or gamma corrected?) RGB values
 > via matrix multiplication like so
 > 
 >  R' = 1.4234 * R + -0.3230 * G + 0.0061 * B + -29.7088
 >  G' = 0.0600 * R +  0.9677 * G + 0.0719 * B + -31.3147
 >  B' = 0.1562 * R + -0.5104 * G + 1.4530 * B + -28.9059
 > 
Hi Olaf,
thanks for the hint.  I assumed something like this but was confused
by the fourth column.

I wrote a Lua script which allows me to modify each pixel within a PNM
file.  I first applied the lookup-table (sensor specific) and then the
matrix (film material specific).  It seems that colors are correct
now.  When I apply the matrix first, colors are completely wrong.

I don't know whether the range of values in the fourth column is
[0..255] or [0..99].  Since there is no significant difference between
these values (all values are abt. -30), it's nearly impossible to
determine visually which range is correct.  I assumed [0..255].  

The pictures can still be improved with an external tool like
ImageMagick.  They look better with increased values of brightness and
contrast, at least on screen.

Thanks again,
  Reinhard

-- 
------------------------------------------------------------------
Reinhard Kotucha                            Phone: +49-511-3373112
Marschnerstr. 25
D-30167 Hannover                    mailto:reinhard.kotucha at web.de
------------------------------------------------------------------



More information about the sane-devel mailing list