[time-nuts] Scanning darkly

Jack Hudler jack at hudler.org
Mon May 14 01:59:28 EDT 2007


Yes on this type of document (where the screen angle is 45 degrees),
LPI=DPI. Sorry for the confusion; printer speak.
A halftone image can contain dot's of varying sizes or perceived shades of
grey (look at news paper image under magnification).
Depending on the plate, press and paper a halftone dot can be as large as
(on a 133 LPI) .0075" to as small as .001" with 64 sizes in between. 
I hope this makes sense; When scanning a halftone image (where the dot size
varies), to a continuous color device (a square pixel doesn't), you have to
over sample (LPI*2=scan DPI) the line to retrieve the information for single
halftone dot. This dot may be spread across at most 4 pixels. A Gaussian
blur then combines information from those 4 pixels in to 1 and so on.
It can be hard reading an image from one model that was meant to fool your
eye only to move it into another, only to fool yourself again. :)

I noticed that your Epson 3170 outputs 16 bit grayscale! Use it! 

I just hope you dealing with round halftone dots and not something fancy.

-----Original Message-----
From: time-nuts-bounces at febo.com [mailto:time-nuts-bounces at febo.com] On
Behalf Of WB6BNQ
Sent: Monday, May 14, 2007 12:00 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] Scanning darkly

Hi Jack,

Is DPI the same as LPI ?  I have always been confused on that point.  If
they
are the same, why are there two seemingly different specs used for the same
thing ?

I can do TIFF.  How does that compare to the PNG ?

thanks,

Bill....WB6BNQ


Jack Hudler wrote:

>         Try scanning at 300 DPI grayscale (400 max, anything more is a
> waste). Do not use the histogram or descreening functions in your scanner
> software unless you spent some serious dollars. All "descreening" really
> does is increase the actual scanning resolution up to 2 times what you
asked
> for (or fake it), then do some post processing (which you have no control
> over), in order to give you the image you hope you asked for. I prefer to
> get the raw data and do my own post processing because it can be different
> from page to page, and is never the same from manual to manual. Besides
the
> guys that write the scanner software that come with your home scanner;
don't
> really know what they are doing.
>
>         Why 300 DPI? Most manuals are printed at 135 LPI (or 150)
therefore
> scanning at 300 DPI satisfies Nyquist–Shannon (sampling theorem >2*LPI)
> which reduces halftone moiré.
>
>         Do not use lossy image compression such as jpg. Lossy image
> compression screws up the spatial frequency of halftone images, making it
> almost impossible to do any effective post processing. I use PNG because
it
> supports 16 bit grayscale but most scanners won't give you that and
acrobat
> really likes it.
>
>         Moiré can be eliminated after scanning by using a Gaussian blur of
> ~1.5 pixels (see Photoshop). Some scanner software actually do this as
part
> of "descreening" and some high end software ($$$$) may do Fourier analysis
> to calculate the correct Gaussian distribution.
>         Failure to do this step will just reintroduce the Moiré when you
> later downsample the image.
>
>         At this point you want to do a histogram stretch chopping off the
> highs and lows as needed to remove noise. If you have any version of
> Photoshop 6 or greater this can be set up as a batch process.
>
>         Normally I downsample all final text to 150 and leave the
schematics
> at 300 or higher. If you want to OCR then do not downsample prior to OCR.
>
>         If you're plagued by bleed through then you may have to use a
duplex
> scanner or get aggressive in the histogram phase.
>
> Have fun!
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts


_______________________________________________
time-nuts mailing list
time-nuts at febo.com
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts





More information about the time-nuts mailing list