Filmscanners mailing list archive (firstname.lastname@example.org)
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: filmscanners: Pixels per inch vs DPI
> From: Austin Franklin <email@example.com>
> > I didn't leave anything out...it doesn't matter WHAT size pixel, a
> pixel is
> > but a single value of tonality, period. A pixel does NOT contain
> the same
> > amount of information as A dye cloud. As I said, dye clouds are
> variable in
> > shape, and a pixel is only a square (or some fixed shape), and the
> data in A
> > pixel (he said "A") does not represent this shape.
> I dare say an individual dye cloud varies in tone as well as shape, if
> not color as well.
Yes, that is very true...
> There are contaminants you know. There is
> obviously no way one pixel could represent an individual dye cloud
> with complete accuracy. One need not be an engineer to understand
Glad to hear that ;-)
> A more interesting question (to me) would be how many pixels
> are needed to do the job (per cloud, of course).
That number varies a LOT (dye cloud size varies a lot too)...you can always
sample analog at higher and higher frequencies (as in line pair per mm), but
there is a point where the data really isn't useful...at least in a lot of
Assuming a 3 micron "spot" size...
A micron is one millionth of a meter, and a meter is 39.37"...so 3 microns
is 0.00011811 inches. (39.37" * .000001 * 3)...which is 8,466/inch...
A 35mm sensor is, say, 24 x 36, or 1" x 1.5"...and you need 2x to sample, so
1" x 8,466/inch x 1.5" x 8,466/inch = 108M "spots" BUT...camera sensors need
(three) four sensors (RGBG, extra G for contrast) for every pixel, since a
single sensor can only capture one color...so that comes out to 324M sensors
to capture the same information (to some degree) as does a 35mm color film
with 3u spot size...
Note, when a digital camera claims 6M pixels...that's in fact a flat out
lie. It is REALLY 1.5M pixels, with four sensors per pixel...a pixel IS
made up of all three RGB components, so it is really misleading to make the
claims they do. They would be more honest to call it a 6M SENSOR array.
How they get 6M pixel OUTPUT is interpolation...which, of course, means that
%75 of the image data is just made up, and not real image data (to a large
degree). And you thought the scanner dynamic range issue was misleading...
And, because you need four sensors to sample one spot of the same size, the
sensors would have to be 1.5u x 1.5u in size.
Now, let's take this from another standpoint...film resolution. Let's
assume 200lp/mm (which Royal Gold 25 does). That's 400 lines/mm (a line
pair is two lines, one black and one white). 25.4 mm per inch, so 25.4
mm/inch x 400 lines/mm = 10,160 lines/inch. 1" x 10,160 x 1.5" x 10,160 =
155M "spots"...but then multiply by 4 because you need four sensors to make
one pixel...or 466M sensors...
There are linear CCDs used in scanners, that are 10k samples across...but,
they aren't 1" wide. The sensor area is more like 3" wide, and the fact
that it's linear, means the wires have a LOT more area to come out of the
sensor than in a packed array, and also a lot less distance to travel to get
to the amplifiers. You really can't compare linear CCDs to one shot digital
camera arrays, they are entirely different animals from a technology and
All the above is purely to replicate with a digital sensor array, the same
amount of information (to some degree) that film has. This is an entirely
different problem than scanning film...since to scan film you need to scan
at 2x the maximum frequency you want to capture, so you would multiply each
dimension by 4...so to really scan "every" dye cloud on a 200lp/mm piece of
35mm film, you'd get a total of near 2G bytes when done!