Archived posting to the Leica Users Group, 2000/10/30

[Author Prev] [Author Next] [Thread Prev] [Thread Next] [Author Index] [Topic Index] [Home] [Search]

Subject: Re: [Leica] Film megabytes and some curious contradictions
From: Johnny Deadman <john@pinkheadedbug.com>
Date: Mon, 30 Oct 2000 17:11:17 -0500

on 30/10/00 3:57 pm, Erwin Puts at imxputs@knoware.nl wrote:

> Look at the facts: I noted earlier  that you should aim for the 135MB
> possible with Kodachrome or even better hi-res BW film. I am shot dead for
> this proposal, as it exemplifies The Idiots Approach To Ridiculous HiRes
> Photography.
> Then I note that maybe 20Mb would be a sensible compromise and I am again
> shot dead because film can handle 100Mb or more and now it exemplifies The
> Idiots Approach to Ridiculous Claims by Digital Photography.
> I am confused!
> Can anyone explain?

No one ever called you an idiot, Erwin. You're way too sensitive about this.

As to the previous numbers I can't comment. But your theoretical
calculations are just way off what actually happens when you scan and print
an image from a negative. First, the grain IS part of the photographic
image. If I shoot 400 film I want to see the grain.

Second, most modern inkjet printers require at least 300 ppi to image at
their highest quality and many people claim there is a visible difference
between 300 ppi and 600 ppi. But let's run with 300 ppi.

Now let's assume you're printing on a widely available printer like an Epson
1160 or an Epson 1270 which will print on paper up to 13x19. So we're going
to use all of that with an inch margin ... so we'll print 17 x 11 roughly.
We want it to approximate the quality of a conventional silver gelatin
print, right, so we NEED to see the grain at that size unless we're just
going to not print our 400 negs that big.

Now let's do the math. 11x17x300x300 = 16 Mb for a monochrome image or 48 Mb
for a color image. If you go to Digital Silver you will find that this is
really the minumum most hybrid photographers would regard as  workable for
high quality output... and decidedly marginal at that.

Now that's just for starters. Like I say, many inkdot-sniffers will claim
you should print at 600 dpi not 300 dpi for maximum quality. Moreover, most
filmscanners will return 12 bits of data, which very useful when, for
instance, you darken a sky in photoshop... you often find your detail in
those last 4 bits. I know I do. So you need 2 bytes not 1 for each pixel.
When you do this calculation you get much bigger numbers: 11x17x600x600x2 =
134 Mb for a monochrome file, three times that for color.

Now I think that is extreme but from personal experience I can tell you that
the difference between an 11x17 print from a 2400 dpi 8-bit scan (already
bigger than the 20 Mb you keep talking about) and an 11x17 print from a 4000
dpi 16-bit scan is ENORMOUS. If this was not the case then there would, for
example, be no market for the Polaroid Sprintscan 4000 scanner.

Either I'm lying or 20 Mb isn't enough.

Your numbers remind me of the oft-quoted example of the aeronautical
engineer and the bee. You may discount my and others' assertions but we have
the advantage of actually producing high-quality digital prints ourselves,
so ours is a practical rather than a theoretical standpoint.

Personally I think the flaw in your reasoning is dazzlingly obvious but my
counterrexample will have to do for the moment.

- --
John Brownlow

http://www.pinkheadedbug.com