[sane-devel] Re: bug in XSANE?

Levente NOVÁK lnovak@dragon.klte.hu
Sun, 22 Jul 2001 20:28:08 +0200 (CEST)

On Mon, 16 Jul 2001, Karsten Festag wrote:

> Hi Oliver,
> O.k., I didn't thought enough what happens to the data if one applies a
> function to it... I think I understood it now. But: even if I reset the
> color corrections directly before the scan, the histogram has its
> spikes+holes. But there is no visible difference between an image
> scanned by xscanimage and xsane, both with linear transfer function. So
> I won't complain.
> Regards
> Karsten
> Oliver Rauch wrote:
> > 
> > when you set color corrections to default value:
> > gamma=1.0, brightness=0, contrast=0 before you do
> > the scan/preview scan the histogram should not
> > have any spikes/holes.
> > 
> > When you use other color correction values with xsane
> > or define a custom gamma table with xscanimage
> > you will get the spikes/holes.
> >

I can confirm Karsten's observation: the spikes and holes _do_ exist when
scanning with Xsane and do not when using scanimage or xscanimage, when
everything is reset at the default values (=no frontend correction of
scanned data). And even if visually the images are very similar, I
consider this a bug. The eyes are simply not able to discern some missing
colour tonalities, but the histogram should be smooth. E.g. for
densitometric work, a histogram with holes and spikes is barely useable
(yes, a modern scanner, although not as good as a real densitometer, can
be used to such purposes).