"Bart van der Wolf" wrote:

It's also quite soft: some number of times softer than a dSLR
original would be taken with a similar quality (US$700 or so prime)
lens.

I'm a bit disappointed by such a cheap 'argument'. Although we both
don't have a 1Ds (Mark II) to prove or disprove such a statement, I do
expect something more to substantiate that claim (comparable image,
web-link, scientific paper, experiment, any proof).

Sheesh. You're one fussy bloke. Oh, Alright. I'll do the work. Sheesh.

I don't have a 1DsM2, but I do have a camera (300D) with roughly the same
pixel pitch and the power pole is still here, so here goes. Check my math
(I'm cropping everything to a 1:1.414 aspect ratio):

1DsM2 = 3328 x 4714 (cropped to 24x34)

Since 35mm = 24x34 mm = (4000 x 24/25.4) x (1.414 x 4000 x 24/25.4)

35mm = 3780 x 5340. (Nikon 8000 pixels)

645 = 39x56 mm = 6140 x 8818 (Nikon 8000 pixels)

So upsampling a 300D shot with a 22mm lens by 1.85x would give the same
pixel magnification as a 4000 dpi scan of 645.

http://www.pbase.com/davidjl/image/34473562/large

And upsampling a 300D shot with a 35mm lens by 1.15x would give the same
pixel magnification as a 4000 dpi scan of 35mm with a 35mm lens.

http://www.pbase.com/davidjl/image/34473670/large

(Click original to compare, obviously. The digital shots were handheld with
300D + 17-40/4.0 at f/8 at 22 and 35mm focal lengths.)

Anyway, we'll argue as to whether there's a significant difference between
the simulated 1DsM2 (300D with the 17-40 at 22m (35mm in 645 = 22mm in
35mm)) and 645 + Tech Pan. (It looks pretty close to me: the grain in the
Tech Pan makes it seem to have more detail than it actually does.)

But everyone will admit that in the simulated 1DsM2 vs. simulated 35mm
comparison, digital kicks film's butt something fierce.

Anyway, my theory was that 16MP 24x36 dSLR would kick 35mm's butt and be
about equal to 645, and that's what I'm seeing here.

David J. Littleboy
Tokyo, Japan