[filmscanners] Re: Advice on scanner settings

["Arthur Entlich" <artistik@shaw.ca>]

I just did a little empirical experiment, to ty to get some idea about
jpeg compression.

I took a rather complex image (a photograph of a very detailed painting)
which I had saved as a TIFF file.  It was from a Kodachrome 25 slide
which I scanned at 14 bit (16 bit in Photoshop) at 4000 dpi.  That
created a 62M file.

I then converted it to a 8 bit file, which brought it down to about
31M.  I then converted it to a jpeg at the highest quality Photoshop
offered (in version 5.5) which was scale12 in baseline standard mode.
It was reduced to about 11M.

Now, here I'll show my ignorance.  I then took the original TIFF image
and copied the jpeg as a layer on top of it, and  I then selected
"difference" as the blend relationship between the background tiff and
the layer one jpeg. The image went completely black even at very high
zoom, I only see black.  I then tried this with higher jpeg compressions
and yet it is still all black.

So, I am confused.  I presumed that using the difference blend
relationship between the background and top layer would show any pixels
which had different colors (I believe difference subtracts the RGB
values from each other) and yet I see nothing but what appears to be a
pretty pure black, even with jpeg scale 1.

OK, I just did another test, and I see it takes quite a bit of change in
the layer (playing with hue, contrast, levels, etc) before the
difference blend shows anything meaning, so that is not a good testing
procedure.

Anyone have a good idea how to check two images for changes against one
another such that hue, color, contrast, brightness or any value change
to a pixel would show up clearly as a changed pixel when comparing two
images on top of one another?   I would like to see a quantitative
visual indication of each pixel that is altered by a certain jpeg
setting relative to the non-jpegged tiff.

For instance, I am wondering if jpeg 12 on the Photoshop scale is near
lossless or not.  A jpeg 12 in Photoshop seems to be a 3:1 compression,
but I am not sure how much damage it is doing.

Art


LAURIE SOLOMON wrote:
>> However, very high quality JPEG can reduce it to about 8 megs or less,
>>
> with little loss,
>
> I have to disagree with my friend Art on this one.  I do not think that a
> high quality .jpg will reduce the file size significantly more than using
> the lossless LWZ compression with a TIFF file and is less open to artifacts
> and other issues that can arise from lossy compression used by JPEG if and
> when there are numerous iterations of opening and closing of the file with
> changes to it being made with the various iterations.  However, to get to
> the amounts of compression Art has suggested, you would have to save the
> file as a JPEG with significiant amounts of compression at lower quality
> levels with lossy compression.  This increases the possibilities of JPEG
> artifacts and data loss significantly and is not really desirable for
> archiving purposes.
>
> -----Original Message-----
> From: filmscanners_owner@halftone.co.uk
> [mailto:filmscanners_owner@halftone.co.uk] On Behalf Of Arthur Entlich
> Sent: Thursday, February 26, 2009 2:18 AM
> To: laurie@advancenet.net
> Subject: [filmscanners] Re: Advice on scanner settings
>
> All your setting sound reasonable assuming you are wishing to maintain
> the files as a permanent collection of your images in a digital format,
> to be accessed later as needed.  That way you won't need to return to
> the slide except for very specialized purposes.
>
> I'm assuming your collection is going to be quite large, and that it
> extends through many years and many emulsion types.   Multiple passes
> adds a lot of time to the scanning, so you may want to consider if that
> is necessary for all the slides or if some exposures, ISOs, or emulsion
> types don;t require as many passes.  Multi-pass does improve the dynamic
> range and lowers noise artifacts, particularly with dense slide with
> deep shadows, but it may not be necessary for slides with fairly even
> exposure and which are brighter.   I would suggest doing a few tests and
> determine which category of slides need extra passes, as many may not
> (check both the type of scene/exposure and film stock (Ektachrome,
> Kodachrome, etc.) including film ISO, and when the manufacturers made
> large changes in film emulsion. Then you may be able to categorize them
> and determine some can get away with one or two passes, and others would
> benefit from 3 or 4.
>
> Storage involves several issues.  If you are scanning beyond 8 bits per
> color at 4000 dpi, the end result can be up to 100 megs per image.  Even
> at 8 bit, you are talking about 35-45megs in TIFF format.  However, very
> high quality JPEG can reduce it to about 8 megs or less, with little
> loss, but conversion will add a bit of processing time.  If you have
> enough storage you may want to stay with TIFF.  Be sure to back up
> regularly since the scanning is the big time consumer and you wouldn't
> want to lose your images.
>
> As has been mentioned, infrared scans of some Kodachrome films will not
> effectively remove dust and scratches, because there is residue silver
> left in the film even after processing which is opaque to IR, so the
> software cannot tell the difference between the silver and dust.
>
> Good luck.  It's a large project you are tasking on, so anywhere you can
> find efficiencies is worthwhile to take advantage of.
>
> Art
>
>
> Carlisle Landel wrote:
>
>> Bunch,
>>
>> I about to begin scanning a lifetime of slides (mostly Ektachrome but
>> a smattering of Kodachrome) using a Nikon LS-5000 and Vuescan.
>>
>> Are the following settings appropriate?  Why or why not?
>>
>> I'm planning on 4000 dpi for maximum resolution, with 3 samples and
>> the color analog gain set at 1 for all colors.
>>
>> I'm also planning a light infrared screen with no other filtering
>> with respect to colors, grain reduction, or sharpness.
>>
>> I'm planning to auto balance colors using the default options and
>> appropriate slide types.
>>
>> With respect to output, I gather that TIFF is better than JPEG,
>> because JPEG is compressed.  Is that right?
>>
>> Thanks for your input,
>>
>> Carlisle
>>
>> --who figures he'll start scanning now, then figure out how to
>> manipulate scanned images later.
>>
>>
>>
>>
>>
>>
>>
>
> ----------------------------------------------------------------------------
> ------------
> 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