Best way to remove halftone artifacts

Hi,

What's the best way to remove halftone artifacts when
scanning sources with halftone?

Thanks.

Bill

On Mon, 2 Nov 2009 15:26:08 -0800 (PST), Bill <>
wrote:

>Hi,
>
>What's the best way to remove halftone artifacts when
>scanning sources with halftone?

http://scantips.com/basics06.html

me

On Nov 3, 7:39*am, me <m...@mine.net> wrote:
> On Mon, 2 Nov 2009 15:26:08 -0800 (PST), Bill <bulba...@gmail.com>
> wrote:
>
> >Hi,
>
> >What's the best way to remove halftone artifacts when
> >scanning sources with halftone?
>
> http://scantips.com/basics06.html

Those are for older scanners. New scanners have 2400-4800
dpi. In a typical halftone printing, what dpi do they use? Say it is
2400 dpi, then if you match it with a similar 2400 dpi scanner
mode and you get one on one correspondence, the moire
pattern can be minimize or gone?

Bill

Bill <> wrote:

> On Nov 3, 7:39*am, me <m...@mine.net> wrote:
>> On Mon, 2 Nov 2009 15:26:08 -0800 (PST), Bill <bulba...@gmail.com>
>> wrote:
>>
>> >Hi,
>>
>> >What's the best way to remove halftone artifacts when scanning
>> >sources with halftone?
>>
>> http://scantips.com/basics06.html
>
> Those are for older scanners. New scanners have 2400-4800
> dpi. In a typical halftone printing, what dpi do they use? Say it is
> 2400 dpi, then if you match it with a similar 2400 dpi scanner
> mode and you get one on one correspondence, the moire
> pattern can be minimize or gone?

The advice would still be valid except IMHO that advice page is not well
written, it seems very muddled to me.

Essentially to avoid halftone artefacts you need to scan at a much higher
resolution than the dot screen (to avoid moire) and then resample the image
down to the same resolution as the dot screen, which is the highest
resolution at which there is useful picture information. Before this second
step it helps to blur the image enough to make the dots run together to
prevent them being preserved as spurious detail during downsampling. If the
scanner has a descreen filter you simply tell it the dpi of the dotscreen
and it will do the rest, though IME you can get better results by doing the
descreening yourself.

Dotscreen pitch depends on the type of publication and possibly on country
too. If you have a decent lupe and a ruler you can measure it yourself
directly. IME modern UK newspapers typically use 100 dpi, magazines 150 dpi
and fine art books about 220 dpi.

Suppose the dpi is 150, you could scan at 1200 dpi (giving 8 pixels per
dot) and then apply a blur of up to 8 pixels diameter (ie. radius 4 blur if
using photoshop etc) which reduces the effective resolution back to 150dpi,
blending the dots together. (You might want to try less blur than this
though to prevent losing any detail.) Now resample down to an actual
150dpi. You don't have to do this last bit, but it's all the information
there is anyway so you'd be wasting space keeping a higher resolution
unless you need a higher resolution file for printing purposes.




--

Gordon Freeman

On Tue, 3 Nov 2009 17:40:11 +0000 (UTC), Gordon Freeman
<> wrote:

>Bill <> wrote:
>
>> On Nov 3, 7:39*am, me <m...@mine.net> wrote:
>>> On Mon, 2 Nov 2009 15:26:08 -0800 (PST), Bill <bulba...@gmail.com>
>>> wrote:
>>>
>>> >Hi,
>>>
>>> >What's the best way to remove halftone artifacts when scanning
>>> >sources with halftone?
>>>
>>> http://scantips.com/basics06.html
>>
>> Those are for older scanners. New scanners have 2400-4800
>> dpi. In a typical halftone printing, what dpi do they use? Say it is
>> 2400 dpi, then if you match it with a similar 2400 dpi scanner
>> mode and you get one on one correspondence, the moire
>> pattern can be minimize or gone?
>
>The advice would still be valid except IMHO that advice page is not well
>written, it seems very muddled to me.
>
>Essentially to avoid halftone artefacts you need to scan at a much higher
>resolution than the dot screen (to avoid moire) and then resample the image
>down to the same resolution as the dot screen, which is the highest
>resolution at which there is useful picture information. Before this second
>step it helps to blur the image enough to make the dots run together to
>prevent them being preserved as spurious detail during downsampling. If the
>scanner has a descreen filter you simply tell it the dpi of the dotscreen
>and it will do the rest, though IME you can get better results by doing the
>descreening yourself.
>
>Dotscreen pitch depends on the type of publication and possibly on country
>too. If you have a decent lupe and a ruler you can measure it yourself
>directly. IME modern UK newspapers typically use 100 dpi, magazines 150 dpi
>and fine art books about 220 dpi.
>
>Suppose the dpi is 150, you could scan at 1200 dpi (giving 8 pixels per
>dot) and then apply a blur of up to 8 pixels diameter (ie. radius 4 blur if
>using photoshop etc) which reduces the effective resolution back to 150dpi,
>blending the dots together. (You might want to try less blur than this
>though to prevent losing any detail.) Now resample down to an actual
>150dpi. You don't have to do this last bit, but it's all the information
>there is anyway so you'd be wasting space keeping a higher resolution
>unless you need a higher resolution file for printing purposes.

My editor, Photoline, has a descreening filter built in. You adjust the
size, intensity, and threshold sliders; along with what color-space you
want to work in, RGB, HIS, or Lab. Works much better than doing it manually
with gaussian blurring and deconvolving to reclaim details. I'm not sure
what process that filter uses but it works better than anything I've tried
to do manually.

Data-Point

On Nov 4, 1:40*am, Gordon Freeman <Gor...@freeman.invalid> wrote:
> Bill <bulba...@gmail.com> wrote:
> > On Nov 3, 7:39*am, me <m...@mine.net> wrote:
> >> On Mon, 2 Nov 2009 15:26:08 -0800 (PST), Bill <bulba...@gmail.com>
> >> wrote:
>
> >> >Hi,
>
> >> >What's the best way to remove halftone artifacts when scanning
> >> >sources with halftone?
>
> >>http://scantips.com/basics06.html
>
> > Those are for older scanners. New scanners have 2400-4800
> > dpi. In a typical halftone printing, what dpi do they use? Say it is
> > 2400 dpi, then if you match it with a similar 2400 dpi scanner
> > mode and you get one on one correspondence, the moire
> > pattern can be minimize or gone?
>
> The advice would still be valid except IMHO that advice page is not well
> written, it seems very muddled to me.
>
> Essentially to avoid halftone artefacts you need to scan at a much higher
> resolution than the dot screen (to avoid moire) and then resample the image
> down to the same resolution as the dot screen, which is the highest
> resolution at which there is useful picture information. Before this second
> step it helps to blur the image enough to make the dots run together to
> prevent them being preserved as spurious detail during downsampling. If the
> scanner has a descreen filter you simply tell it the dpi of the dotscreen
> and it will do the rest, though IME you can get better results by doing the
> descreening yourself.
>
> Dotscreen pitch depends on the type of publication and possibly on country
> too. If you have a decent lupe and a ruler you can measure it yourself
> directly. IME modern UK newspapers typically use 100 dpi, magazines 150 dpi
> and fine art books about 220 dpi.

how can you use a loupe and ruler to measure the halfttone
dots dpi (or lpi) when the dots are of varying sizes?? If the
portion of the picture is of lighter shade, there are less dots.
If more dark, more dots.. so which of them should you base
the dpi (or lpi) measurements?

Bill

>
> Suppose the dpi is 150, you could scan at 1200 dpi (giving 8 pixels per
> dot) and then apply a blur of up to 8 pixels diameter (ie. radius 4 blur if
> using photoshop etc) which reduces the effective resolution back to 150dpi,
> blending the dots together. (You might want to try less blur than this
> though to prevent losing any detail.) Now resample down to an actual
> 150dpi. You don't have to do this last bit, but it's all the information
> there is anyway so you'd be wasting space keeping a higher resolution
> unless you need a higher resolution file for printing purposes.
>
> --- Hide quoted text -
>
> - Show quoted text -

Bill

Bill <> wrote:

>> Dotscreen pitch depends on the type of publication and possibly on
>> countr y too. If you have a decent lupe and a ruler you can measure
>> it yourself directly. IME modern UK newspapers typically use 100 dpi,
>> magazines 150 d pi and fine art books about 220 dpi.
>
> how can you use a loupe and ruler to measure the halfttone
> dots dpi (or lpi) when the dots are of varying sizes?? If the
> portion of the picture is of lighter shade, there are less dots.
> If more dark, more dots.. so which of them should you base
> the dpi (or lpi) measurements?

Halftone dots vary in size but their spacing is fixed. Maybe you're
thinking of inkjet prints where the dots are fixed size but their
spacing varies. With a printing press the different tones are created by
the dots taking up a bigger or smaller proportion of the fixed space
allotted to them.

I suppose it's possible that their may be printing presses that use
random dot dithering like an inkjet but I've not noticed any
publications that do this. I'm not sure if there are any technical
reasons for not doing it that way, but I suspect the need for very small
closely spaced dots might cause inking problems with the printing plates
bearing in mind that printing ink is quite thick.

(Incidentally if you are scanning inkjet prints and not proper halftone
stuff then the issue of moire shouldn't arise since inkjet dots are
randomly spaced so there will be no regular pattern to clash with the
scanner pixels.)

Gordon Freeman

On Nov 4, 9:00*am, Gordon Freeman <Gor...@freeman.invalid> wrote:
> Bill <bulba...@gmail.com> wrote:
> >> Dotscreen pitch depends on the type of publication and possibly on
> >> countr y too. If you have a decent lupe and a ruler you can measure
> >> it yourself directly. IME modern UK newspapers typically use 100 dpi,
> >> magazines 150 d pi and fine art books about 220 dpi.
>
> > how can you use a loupe and ruler to measure the halfttone
> > dots dpi (or lpi) when the dots are of varying sizes?? If the
> > portion of the picture is of lighter shade, there are less dots.
> > If more dark, more dots.. so which of them should you base
> > the dpi (or lpi) measurements?
>
> Halftone dots vary in size but their spacing is fixed. Maybe you're
> thinking of inkjet prints where the dots are fixed size but their
> spacing varies. With a printing press the different tones are created by
> the dots taking up a bigger or smaller proportion of the fixed space
> allotted to them.
>

So how do you measure the fixed spacing by a ruler and loupe,
maybe by looking at the darkest part of the image? But when I
do, I can see alternating white and black dots. I wonder if
one halftone cell is made white and one halftone cell black or
the dots in the halftone cell were made half size black. The
source I was looking at (experimenting) was a black and
white picture in a book published in 1983 in gross paper.
Before the advent of digital halftoning, I wonder what is the
equivalent output resolution dpi they use? 600 dpi? 1200?
2400 dpi? any ideas?

Bill


> I suppose it's possible that their may be printing presses that use
> random dot dithering like an inkjet but I've not noticed any
> publications that do this. I'm not sure if there are any technical
> reasons for not doing it that way, but I suspect the need for very small
> closely spaced dots might cause inking problems with the printing plates
> bearing in mind that printing ink is quite thick.
>
> (Incidentally if you are scanning inkjet prints and not proper halftone
> stuff then the issue of moire shouldn't arise since inkjet dots are
> randomly spaced so there will be no regular pattern to clash with the
> scanner pixels.)

Bill

Bill wrote:
> On Nov 4, 1:40 am, Gordon Freeman <Gor...@freeman.invalid> wrote:
[...]
>> Dotscreen pitch depends on the type of publication and possibly on country
>> too. If you have a decent lupe and a ruler you can measure it yourself
>> directly. IME modern UK newspapers typically use 100 dpi, magazines 150 dpi
>> and fine art books about 220 dpi.
>
> how can you use a loupe and ruler to measure the halfttone
> dots dpi (or lpi) when the dots are of varying sizes?? If the
> portion of the picture is of lighter shade, there are less dots.
> If more dark, more dots..

No, the number of dots stays the same - it's the size of the dots that
varies.


--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------

Bob Larter

Gordon Freeman wrote:
> Bill <> wrote:
>
>>> Dotscreen pitch depends on the type of publication and possibly on
>>> countr y too. If you have a decent lupe and a ruler you can measure
>>> it yourself directly. IME modern UK newspapers typically use 100 dpi,
>>> magazines 150 d pi and fine art books about 220 dpi.
>> how can you use a loupe and ruler to measure the halfttone
>> dots dpi (or lpi) when the dots are of varying sizes?? If the
>> portion of the picture is of lighter shade, there are less dots.
>> If more dark, more dots.. so which of them should you base
>> the dpi (or lpi) measurements?
>
> Halftone dots vary in size but their spacing is fixed. Maybe you're
> thinking of inkjet prints where the dots are fixed size but their
> spacing varies. With a printing press the different tones are created by
> the dots taking up a bigger or smaller proportion of the fixed space
> allotted to them.
>
> I suppose it's possible that their may be printing presses that use
> random dot dithering like an inkjet but I've not noticed any
> publications that do this.

I've seen it in Japanese magazines.

--
W
. | ,. w , "Some people are alive only because
\|/ \|/ it is illegal to kill them." Perna condita delenda est
---^----^---------------------------------------------------------------

Bob Larter