re:Digital vs Analog - Part 1

[P3D Larry Berlin <lberlin@xxxxxxxxx>]

>Date: Thu, 12 Dec 96 14:31:18 PST
>From: P3D Michael Kersenbrock comments:
>
>>  (Gregory J. Wageman quote)This can't be right (about film, I mean).  In
order for film to behave
>> as you say (with light-darkened silver halide crystals acting like a
>> halftone), then adjacent crystals would have to have knowledge of what
>> their neighbor is doing................snip................
>
>Why?  Does an acorn have to understand cell division to grow a leaf?  If
>it acts like it's halftoning, then it is, whether it knows it or not.
>The CCD cell in a digital camera is an "analog" device, so why isn't
>that digital camera called "analog"? 

*************  Greg's message seemed to indicate that silver halide crystals
individually develop varying densities of black so they are not binary
equivalents. In a digital camera, it's the way the information is handled
that determines the name *digital*. Each analog packet from the CCD is
digitized as soon as it's read so only the sensor is analog. Even the
sampling method of the device is a digital process.

>
>>(Gregory J. Wageman quote)
>> Like digital imaging companies don't have a vested interest in convincing
>> a skeptical photographic community that film resolution really isn't as high
>> as it is?  A reasonable estimate of the resolving power of film is 100 line
>> pairs per millimeter.  These numbers come from the film companies themselves.
>> By simple math, this is 200 lines/mm, giving a 33mmx23mm 35mm film chip
>> the equivalent of about 6600x4600 pixels, which is a little over 30
>> million pixels.  I'm sure they'd love you to believe it's only a tenth
>> of that.
>
>You are not addressing that of which I wrote.  Each pixels in a common
>digital image contains 24-bits of information.  I don't see that to 
>be true in the film's case in the way measured.

***********  I liked Michael K's description but it isn't based on multiple
shades within the crystals. If *shaded crystals* is the true state in film
emulsion, you would have to assign the 24 bits to each grain in the
emulsion, which is a pretty high resolution however you want to measure it.
Added to the fact that if they are shaded, they will not be limited to a
finite 24 bits of resolution, though at that microscopic level I dare say it
would be close to impossible to resolve by eye, differences between 24 bit
or 500 bit.

>
>With digital, each pixel can be completely stark-black or completely 100% white
>at full resolution .... or some exact value inbetween.  I don't believe
that is 
>true for each line in your resolution test as it is usually measured (correct
>me if I'm wrong on that).

************  A big advantage in film is the fact that the resolving bits,
or crystals are NOT in neatly ordered rows and columns. The fact that
digital is based on such neat rows and columns works against the resolution
factor in digital imaging. It's necessary to synthesize the inbetween
locations by differences in shading of the surrounding pixels. This means
that each pixel is NOT faithfully registering it's own space, unlike
emulsion crystals. This further dilutes the resolving power of digital
images. The solution to this conventionally is to multiply the number of
rows and colmns so that you have multiple pixels for each of the ones you
previously were considering. The fact is that if you do that to a sufficient
degree, you have whatever resolution you want, although today's CCD devices
can't be so multiplied on the physical level. Within the arena of digital
processing you can develop a digital image with far more resolution than
film but would then be challenged to reproduce it in a suitable medium to
convey all that information.

Theoretically a digital system that could use virtual pixel locations more
randomly distributed might have a resolution advantage over a conventional
digital image with a similar amount of data. But to work it would need some
amount of data to determine how and where each pixel was located. Since
simply multiplying the number of pixels in ordered arrays leaves you with a
handy way of getting to them (manipulations, storage, etc) it seems that it
is still the best solution. We just need the technology to advance to the
next logical level so that it does indeed exceed film processes in terms of
resolution.

I think that the real solution would be to capture multiple images with
existing digital devices and synthesize the in-between resolution based on
the multiple samplings of an existing scene. Multiple captures of
instantaneous information would have far more content than any single
capture (3D has more content than 2D) and be much faster than any multiple
pass system that can only be used with unmoving subjects. Theoretically this
could be done automatically in *real time* and further filtered or enhanced
by hand after the initial capture. 

This would involve special software but could theoretically provide images
1000x the resolution of film. I'm just puzzled by why this hasn't already
been done with digital cameras. I think the basic process is already used
for various image enhancement applications. The apparent lack of movement in
this direction  would be of more concern to me than the *digital company's*
brainwashing potential for getting people to accept a lower resolution
device in place of film. We know digital has the potential so let's go for
it now, why all this *mucking about in hyperspace* (a quote from Douglas
Adam's *Hitchhiker's Guide to the Universe*, which seems to apply here. : -)

A side benefit to this potential development would be ... 

(pause - to be continued in next post... )

Larry Berlin

Email: lberlin@xxxxxxxxx
http://www.sonic.net/~lberlin/
http://3dzine.simplenet.com/


------------------------------