[filmscanners] RE: Suggestions for scanning 4x5 transparencies

["Laurie Solomon" <laurie@advancenet.net>]

Paul,
If the intent of your technical commentary was to explain the reasoning
behind your argument, make me understand the grounds for your points, or to
convince me to accept your position, I am afraid that it missed its mark due
mainly not to the correctness or incorrectness of your conmments but to the
fact that it is beyond my tecninical comprehension.  I know nothing of
things like Nyquist and the like.  Admittedly a limitation on my part but I
claim no science or engineering credentials, training, or interest in
acquiring them.  On technical physics and engineering matters like Nyquist
and things like that I either have to take your word for it on face value or
reserve judgement until such time as I get an elaboration in different more
everyday terms that I can understand.

Bob Frost posted the message on the Epson list that you referred to.  In all
honesty, I did not see it when it was originally posted on that list either
because I missed it in skimming the incoming posts or it was put into
quarintine as spam by my ISP for one reason or another and never transmitted
to me (something which has been happining with some regularity to posts from
Yahoo groups in a random fashion when something in the post (the paid Yahoo
ad or its placement marker) trigger the anti-spam program.

After reading the post, I acknowledge that the resampling initially to 720
ppi/dpi by the printer driver was discussed and that it even appears on the
surface have been proven that the prijtner driver does resample incoming
files to 720 ppi/dpi upon their being input to theprinter  driver, but I
cannot say that it actually was definitely proven beyond a shadow of a doubt
(in that well known courtroom phrasing)that this conversion took place
before any dithering.  I also find that some of the language of Kennedy's
argument are open to some interpretation as it is stated which for me leaves
the question in doubt as to the stage in the printer driver processing that
the 720 ppi resampling takes place; BUT I ACKNOWLEDGE that Kennedy does
agree with you that the Epsson desktop printers do resample to 720 ppi all
incoming files regardless of the resolutions of those files. However, in his
statement, "Every image that you send to the printer is first resampled to
720ppi before being converted into ink dots which are placed on the page at
the dpi setting you have selected," (an example of what I see as ambiguous
if not vague language), it is unclear to me if he is saying this conversion
takes place before or after dithering.  Indeed, I do not think he addresses
the dithering issue at all in his remarks.  This ambiguity and vagueness
also displays itself with respect to if it is better to send the printer
files that have internal resolutions closer to the 720 figure than to send
it files with resolutions that are further from that 720 figure as
illustrated in the following quote:

        If you then resample the image to 240ppi using bicubic resampling then
        you significantly reduce the amplitude of the aliased component in the
        image itself (since bicubic reproduces aliasing with little amplitude).
        Furthermore, when resampled by the printer, there is no alias component
        of this 240ppi image to be FM modulated.

I realize that this statement makes reference to bicubic resampling of the
image prior to sendiing it to the printer; but I think it could be argued
that the same would be the case for images whose original resolutions were
240ppi and involved no resampling prior to being sent to the printer. I also
realize that in his statment as submittted by Bob, Kennedy makes no mention
of sending image files to the printer with resolutions that approach the 720
ppi figure in order to avoid alaising, artifacts, or increase print quality.
I would think that if this were the case or if his post had proven this to
be the case he would have mentioned that the higher the resolution
approaching the 720 ppi mark, the better the printer outcome quality and
print quality.  His statement appears to imply that 240 ppi is good enough
and anything higher does not significantly contribute to print quality; but
hten I might be reading too much into his statements.

Regarding my comments on using even multiples of 720 ppi/dpi, I take
responsibility for my poor choice of wording and articulation which very
well may have colored your response.  You are right in saying that my
illustration appears to be multiples of 30 dpi/ppi.  I did not know of a
better way of articulating what I was attempting to say and would up
slecting misleading language.  Kenedy's post says  what I intended when he
says:
         "Since the Epson printer resamples the image at exactly 720ppi then you
         should, in theory, send it data which is at an integral division of
         720ppi - this means 720ppi, 360ppi, 240ppi, 180ppi etc. since all other
         resolutions can, in theory, produce aliasing.

I hope this adds some clarity to what I was attempting to suggest.  It also
seems to indicate that on this we are closerthan not to being in agreement
and saying the same thing more or less.  But then again, maybe not.

Eating crow would be much easier if it tasted better. :-)

-----Original Message-----
From: filmscanners_owner@halftone.co.uk
[mailto:filmscanners_owner@halftone.co.uk]On Behalf Of Paul D. DeRocco
Sent: Sunday, October 27, 2002 8:41 PM
To: laurie@advancenet.net
Subject: [filmscanners] RE: Suggestions for scanning 4x5 transparencies


> Laurie Solomon wrote:
>
> While I will make no claims to completely comprehending the technical
> arguments and specifics of the engineering nor do I wish to argue
> those things, I do think that your example is a bad one in that it is
> an example of exactly the oppositie of what I said.  I said that the
> multiples of 720 had to be even multiples which would be along the
> lines of the following
> 90,120,150,180,210,240,270,300,330,360,390,420,450,480,510,540,570
> ,600,630,6 60,690....750,780,810,...,1440,..., and 2880 - not uneven
> or odd multiples which 700 dpi would be.  I would even go as far as
> to suggest that the reason that Epson talks of those resolutions like
> 720, 1440, 2880, and 5760 un terms of dpi rather than ppi may be
> precisely because it is in printing terms which does account for
> dithering and stands as such as being equivalent in that regard to
> halftones whose resolutions take into account the line screen
> multiple of the halftone dots in stating halftone resolutions. At any
> rate, it is because you have selected an uneven multiple of the 720
> figure that you wound up with the extra line of black and an extra
> line of white (or double width line on one of the lines of each).  In
> that regard, I believe you example illustrates the reasoning behind
> my statement and supports my contention that it is more frequently
> than not factors such as imputing an uneven multiple of the 720 dpi
> figure that produces the aliasing and artifacts more than the
> closeness of the match between the two numbers.

The numbers you mention all seem to be multiples of 30, but most are not
multiples of 720, so I'm not sure what you're driving at. But what
eliminates all moire is to make the resolution of the image an integral
_submultiple_ of 720. (That way, the spatial harmonics of the image that
exceed the 360lpi Nyquist alias back on top of other harmonics, instead of
to harmonically unrelated frequencies--if anyone's interested.) It isn't
sufficient to make them submultiples of the dot resolution (e.g., 1440 or
2880), since, as I said, the aliasing occurs when the driver resamples to
720ppi, not when it dithers to 1440 or 2880dpi. For instance, if you print
alternating 1-pixel black and white lines at a resolution of exactly
1440ppi, the dumb resampling will simply choose every other pixel, resulting
in either all white or all black, not gray.

--

Ciao,               Paul D. DeRocco
Paul                mailto:pderocco@ix.netcom.com

----------------------------------------------------------------------------
------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title
or body

----------------------------------------------------------------------------------------
Unsubscribe by mail to listserver@halftone.co.uk, with 'unsubscribe 
filmscanners'
or 'unsubscribe filmscanners_digest' (as appropriate) in the message title or 
body