Olympus-OM
[Top] [All Lists]

[OM] Re: Dipping our Toe Into Digital

Subject: [OM] Re: Dipping our Toe Into Digital
From: Joe Gwinn <joegwinn@xxxxxxxxx>
Date: Fri, 20 Dec 2002 16:45:16 -0500
Comments interspersed below.

At 5:59 PM +0000 12/20/02, olympus-digest wrote:
>Date: Fri, 20 Dec 2002 09:58:14 -0800
>From: Jan Steinman <Jan@xxxxxxxxxxxxxx>
>Subject: [OM] Re:  Dipping our Toe Into Digital
>
> >From: Albert <olympus@xxxxxxxxxxx>
> >
> >with the introduction of the Kodak 14Mpx SLR, it would seem like you
> >can get what you never thought possible before, Medium format quality in
> >a 35mm SLR.
>
>That depends on whose numbers you believe. My research indicates that an 
>optimal 35mm frame may contain as much as 18Mpx. Medium format may contain 
>40Mpx.

Reasonable 25mm film cameras resolve something like 50 line pairs per 
millimeter; sometimes better in the center, sometimes less at the edges.  At 
two pixels per line pair, that's 100 pixels per millimeter.  A 35mm frame is 24 
by 36 mm, so we have (24*100)(36*100)= 8.64 million pixels (each having all 
three colors), or (8.64)(3)= 25.92= 26 million pixels (as usually quoted for 
digital cameras).

So, 18 Mpix is a bit low to be "optimal", but it isn't that far off: 26/18= 
1.44 to 1.

Hmm.  I see.  The 18 Mpix appears to assume the typical compromise 
red-green-green-blue pattern, where the R:G:B ratio is 1:2:1, so the effective 
multiplier from film pixels (which are tri-color) to CCD pixels is two, not the 
three one uses for 1:1:1 ratio sensors:  8.64*2= 17.28 million pixels, close 
enough.

As for medium format, a 6x7 image has 3*(60*100)^2= 108 million pixels (as 
camera pixel counts are usually quoted) for 1:1:1, and 72 million pixels for 
1:2:1 ratio.

The Foveon sensor is 1:1:1, while the usual prosumer CCD cameras are 1:2:1.  
The $10K pro studio cameras are mostly 1:1:1, and may have three CCD chips in 
them, just like studio TV cameras.


>In 1997, I predicted that price/performance parity between digital and 35mm 
>was 8 years off. I stand by that: it's now 3 years off.

I assume that you used Moore's Law, that holds that semiconductor technology 
doubles in performance (halves in cost for the same performance) every 18 
months.  Confirmed below.


>(That doesn't mean there won't be earlier performance-parity cameras, just 
>that they won't be affordable. The Kodak is what, a $3,000 body?)

When I bought my first OM-1n body and 50mm f/1.4 lens, it was $300 in 1975, 
which would be $1100 today.  Let's assume that the Kodak body requires a $300 
lens, for a total of $3300.  At the Moore's Law rate, 3.3:1 will take 31 
months, call it 3 years, so your estimate isn't that bad.


>I think price/performance parity with medium format is still 5 years out, with 
>4x5, 8 years, and with 8x10, 11 years.

Yes.  The ratio of film areas is 60^2/(24*36)= 4.167.  It will take 
18*log2(4.167)= 37 months, call it three years more to achieve parity.  The 
total is 31+37= 68 months, or 5.7 years.

Large-format:  4X5 has an image of 3.75 by 4.75 inches, or 
(3.75*25.4)(4.75*25.4)/(24*36)= 13.3 times larger than 35mm, which will take 67 
months, a total of 31+67= 98 months, or 8.2 years.

What will take the longest is movies.  The The Two Towers, the latest Lord of 
the Rings film: three hours of 35mm film.  Movie frames are half-frame size 
(like a Pen) and you get 24 frames per second.  Let's assume each frame is 
24x18, uses 24-bit color, and so contains (24*100)(18*100)(3*1)/(1024^2)= 12.36 
Mbytes.  Three hours at 24 frames a second is (3*60*60)(24)= 259,200 frames, 
and 3,128 Gbytes, or 3 Terabytes.  The proposed new blu-ray DVD disks can store 
something like 30 GB per disk, so it would take more than 100 such disks per 
movie.  Nothing else currently available beats the storage capacity of film, 
and certainly not at the price of film.


>Then kiss film goodbye in ~2013.

No; nothing is ever that clean.  There will be a very gradual transition, 
because people will wait for their current equipment to wear out.  And the 
movie industry will still need vast quantities of film.  One print of The Two 
Towers is (259,200 frames)(18mm)= 4.7 Km long, and there are thousands of 
prints in circulation, so all those theaters can show the film at the same time.

A 36-exposure 35mm film is about 2 meters long, so one such movie print is 
equivalent to 2,333 rolls of 36-exposure 35mm film.


>According to a recent survey, film sales have already peaked; film is now on 
>the downslope of Geoffrey Moore's technology adoption lifecycle, characterized 
>by what Moore calls "technology laggards." (in "Chrossing the Chasm")

It may be that film has peaked or will soon peak, but it's still orders of 
magnitude larger than digital.  I recall someone posted the actual numbers on 
the reflector a while ago.  If I recall, the data came from Kodak's Annual 
Report.

Look at us -- we happily use mechanical cameras from thirty years ago, and 
debate when the right time to go digital might be.  Technology laggards indeed. 
 And proud of it.

For unintentional humour, it's hard to beat the management books published ten 
years before the present -- we believed back then, but in the fullness of time 
all the silly assumptions have been laid bare.


>This is not mere crystal-ball gazing; it's simple back-of-envelope math based 
>on Moore's Law. (different Moore :-)

The killer is that the electronics are only a part of the total cost of a 
camera, and the optical and mechanical components do not follow Moore's Law, 
except that much of the mechanical complexity of cameras has been eliminated: a 
camera today has simple mechanicals controlled by a little computer chip.


>(Of course, there will always be a niche market for fine art photochemistry, 
>just as some brush-media artists still mix their own egg temupra.)

Or develop their own photos?


>So you "fence sitters" still have about 3 years to grouse and whine, or you 
>can start getting some actual experience on a state-of-the-art-minus-one 
>camera TODAY, then step up to the plate in 3 years.

Nah.  It's never too late.  And digital will get better and better, and simpler 
to use as well.  With film, the technology and ergonmetrics are pretty well 
figured out, while we are still figuring out what a digital camera ought to be 
and do.  Present-day cameras (digital and film wunderbricks) are far too 
complex, being encrusted with so many features that nobody can remember how to 
take a picture, and burdened with misfeatures like agonizing shutter lag and 
slow picture storage time andinadequate capacity.  Not to mention battery 
problems.

Joe Gwinn


< This message was delivered via the Olympus Mailing List >
< For questions, mailto:owner-olympus@xxxxxxxxxxxxxxx >
< Web Page: http://Zuiko.sls.bc.ca/swright/olympuslist.html >


<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz