9 Aug
2005
9 Aug
'05
6:40 a.m.
On Mon, 08 Aug 2005 23:47:54 +0200, Tony Duell <ard at p850ug1.demon.co.uk>
wrote:
An excellent lens and high-resolution B&W film may give you up to 100
lp/mm.
The pixel depth will be 12 bits or more.
A 36mm wide negative will accomodate 3600 line pairs or 7200 pixels.
To reliably resolve that digitally without artifacts, the Nyquist theorem
tells us we need at least 14400 pixels.
The equivalent photodetector to match high-resolution film will then be
9600x14400, giving us 138240000 or for short 138 megapixels.
High quality colour film can resolve around 64 lp/mm. This will give us
3072x4608. Assuming a normal lens will work as a low pass filter, we
forget about Nyquist and end up with 14155776 as the lower limit for 35mm
equivalent quality,
I suspect that pixel depth is the biggest problem with current detectors.
When I scan my slides, 3x12 bits per pixel is the minimum to avoid quality
loss. With a scanner, you can easily gain another but by doubling the
number of passes. This does not work with a camera, the detector needs to
have enough depth (signal/noise ratio) to begin with,
--
Bj?rn
Re: "I reckon that a good 35mm camera (good meaning a top-end lens and
fine grain film) is equivalent to 12-20 megapixels."
My understanding from the camera companies, which I've heard more than
once
(and going back to 2000, before any such digital cameras existed) is
that
they consider 35mm film about the equivalent of 6 megapixels. It's not
This may well be the case for a reasonable compact camera with a zoom
lens, and normal colour print film processed in the the local overnight
photo shop (actually, I think 6Mpixels is better than that...). I also
think that a top-end fixed focal length lens, fine grain film, and
careful processing/printing will be a lot better than that.