That image of the car taken on 8 x 10 paper is a classic pinhole look: lots of foreground,
DoF, B&W. The sky reflected in the building windows adds a lot. Somehow, it only
got vignetting at the bottom --- where it helps -- and not at the top where it would
darken the sky. I'm guessing it was cropped?
I'm surpirsed by how good the color images taken with the modified body cap look,
considering DX format and that they are close-ups.
I tried the same thing: made a pinhole bodycap for my Nikon F2 Photomic. The images
were about the same level of unsharpness. as yours. I didn't try any close-ups.
With 35 mm format, I was expecting a lot of diffraciton, as the pinhole is very small.
But I took some more shots with a blue filter to see if it would improve. It didn't, so
I'm thinking the pinhole was actually too large. I think I used Lord Rayleigh's formula.
There seems to be a diversity of opinion about the optimum pinhole size--including in
physics. Most of the formulas amount to:
pinhole_diameter = K * square_root( focal_length * waveflength)
where K is some constant.
Sometimes you see this writtem as:
pinhole_diameter = square_root( K' * focal_length * waveflength)
where K' = square(K).
Nearly everyone uses 555 nm (yellow-green spectral color) as the average of
visible light.
Where they differ is in the constant. There was a post about this on the Photography
forum at Stack Exchange:
https://photo.stackexchange.com/questions/46489/how-to-calculate-the-optimal-pinhole-sizeErick Renner's superb book
Pinhole Photography gave me a clue. He mentioned
a French photographer in thelate 1900s anmed Jules Combe, "whose pinhole photographs
were very sharp" (p. 126) and gives a (very odd) formula. This turns out to be much smaller
than most pinhole sizes that have been suggested over the years.
Building on the post at StackExchange, I worked out a table of contstants K based on
different formulas:
K K*K
------ ------
1.206 1.454 Jules Combe, France, 1899
1.414... 2.000 Josef Petzval, 1957 (square root of 2)
also cited by Wikipedial article "Pinhole Photography"
1.543 2.380 Stanford Pinhole Math (downloaded Dec 28 2018)
1.560 2.440 George Airy (according to PinholeWorks.com)
1.8 3.24 mrpinhole.com Pinhole Size Calculator (results work out at ~1.8)
1.9 3.65 Lord Rayleigh (according to David Balihar)
1.913 3.660 Lord Rayleigh (according to PinholeWorks.com)
I think I've solved the problem. If you'd like to send me a PM with the
distance from your body cap's pinhole to the D300's sensor, I'd be happy
to calculate what I think is the optimum pinhole size (for subjects at infinity).
If you'd also include the diameter of your current pinhole, and the size you typically
display the images (enlargement), I can calculate it's hyperfocal distance,
angle-of-view, image diameter, and aperture number.
If you don't have time, that's fine too.
There is a lot of confusion in the literature about pinholes. For example,
there is "focal length". Pinholes are not converging optics, so they do not
really have a focal length as such.. However, if one choses a minimum
acceptable circle of confusion, then it is possible to calculate a hyperfocal
distance (just as for a lens). Anything beyod this distance will be
acceptably sharp.
Lenses are relatively large, so even at infinity two rays from the subject
can strike the lens at different points. The lens refracts these rays so that they
converge at one point: the focal space (ideally a plane).
But pinhole are much smaller. Only rays that strike the pinhole
get through. And while some of these rays are bent a bit by diffraction
that do not converge (like those rom a zone plate would). By geometry
they would create a circle-of-confusion, by diffraction they create an
Airy disk. What you actually get is a mixture of the two.
As a subject gets closer to a pinhole, the effective size of the pinhole gets
larger, therefore so does the circle-of-confusion. The result is the same
as defocus.
So if one intends to take both close-ups and landscapes, one may need
two different pinholes -- just as you'd need two different lenses -- but for
rather different reasons. There is no "focal plane" with a pinhole--anywhere
you stick the sensor you will get an image, varying only in diameter and
hyperfocal distance.
That image of the car taken on 8 x 10 paper is a c... (