> > What happens or does not happen if you enable quality calibration?
>
> Okay I'll let you know the story. I'm trying to scan a card that's basically
> an embossed white graphic on a slightly off-white background. Well, the
> scanner sees that the thing is almost totally white, and saturates the scan
> really bad. The histogram in xsane does not fall off at the high
> Luminosities, it just saturates. From looking at other stuff on the net,
> this appears abnormal, even for stuff on white backgrounds. I am under the
> ipression that this is what the "Quality Calibration" is for (though I could
> be wrong) is to set the white point so that it is not out of the scanner's
> sensing range, as it apparently is on mine since white and off-white come up
> as the same color -- ffffff white. I would assume it's just an impossible
> scan but someone else did it on another scanner and it came out fine, I'm
> just trying to match that (and not even coming close).
> Now for what happens when when I try to use the calibration. Nothing.
> There is no noticable difference in the scans or the activity performed by
> the scanner. I assume that even if the firmware is doing the calibration
> that it still has to DO something i.e. the lamp has to move prior to the
> actual scan. (Again I could be wrong let me know if I am) In my case I find
> no difference between having the button checked or not in the frontend.
>
No, you are on the wrong way.The quality calibration is not a calibrartion on the
picuture.
The ccd line of a scanner does not have the same output values for each pixel.
So if you would scan a really white area without any calibration, you would
get values - lets say between 220 and 255.
To make all these values unique, the scanner scans one line and calculates a
correction for each ccd pixel. But if you only scan one line you may get a point
that is not white or the lamp does not have the same power over the caliration time
so you get errors in your calibration. In quality calibration it is scanned more than
one line and the values of one ccd pixel are avaraged, so the error of the
calibration
is smaller.
It has nothing to do with you problem!
What you are looking for is higlight and shadow (or black and white point)
definition.
The Astra 1220S only can do this via gamma correction, there are no seperate
functions
for this.
If you are scanning in lineart (black/white) mode, open the standard options window
in
xsane and move the Threshold slider to the right edge. Then do a preview of your
image.
If that does not help, you have to do it in grayscale mode.
Xsane makes available the highlight and shadow function via the hystogram window.
Open the histogram window, between the two histograms there are some colored blocks.
In the recent version there are three sliders in the gray block. Move the mid slider
and
the left slider to the right, as far as possible. Then do a real scan of the image.
On a preview
you will not see any differences, because the preview is always done in 24bpp, the
real
scan uses a 36bpp and afterwards converts it to a 24bpp image.
I hope this works for you!
> Hopefully I'm not totally off base here. I think my scanner is capable of
> doing calibration, since the starting of the driver says
> [umax] f/w support function:
> [umax] ---------------------
> [umax] quality calibration
> but it likely performs it in the firmware, since the
> "change" to firmware 1.5 for this scanner was to add software calibration.
>
> I think I could do better if I had an idea what I was looking for. What is
> the Quality Calibration supposed to work like? Should it happen at the
> beginning of each scan when the button is checked? Should the scan process
> be different? A two pass scan maybe? If it's not working is this because
> the driver is just not telling the scanner to do it properly? Is this
> documented anywhere or is this a trial and error type thing?
>
The difference is in the short time before the real scan begins.
If you don`t see any differences, it is possible that the scanner
always does a quality calibration independent from the quality bit.
May be you see/hear a difference between a preview scan and a
real scan.
Bye
Oliver
-- EMAIL: Oliver.Rauch@Wolfsburg.DE WWW: http://www.wolfsburg.de/~rauch
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail majordomo@mostang.com