Re: Resolution question

"Steve Wolfe" <anx@xxxxxxxxx> wrote:

After getting into the math, it turns out that a C7070 has about 430
pixels/mm,

I was about to argue that your math was wrong, but when I checked, I found
that you were right. Ouch. That really is an insane number of pixels per mm.

so in order to take advantage of that, you'd need to provide resolution
from the lens of at least 220 line pairs/mm.

Ah, but here is something to argue with. Dcams only resolve to about 2/3 of
Nyquist, not all the way out to Nyquist, so you need _decent contrast_ at
150 lp/mm or so. (Check out the test chart images at dpreview.)

That is an astounding figure that you're just not going to achieve. Most
lenses' MTF graphs are only given up to 30 or 40 lp/mm.

And here's something else. It turns out that lens performance goes up as the
format goes down; i.e. it's easier to make a sharp lens for a tiny format
than for a larger format.

(The limit on lens performance is diffraction. Diffraction imposes a 50% MTF
term at approximately 800/(f number), so if you need 150 lp/mm at decent
contrast, you need to be at f/5.6 or wider, and the tiny dcams lenses often
act as though they are close to diffraction limited at f/5.6. They really
are seriously impressive.)

The bottom line is that the insanely tiny insanely high pixel count consumer
dcams actually work, within the laws of physics and optics, quite nicely. As
long as you don't mind restricted dynamic range at the lowest ISO and noise
at higher ISOs.

This is the reason that since the inception of photography, to capture
more detail, the best choice has been a larger film size, and the same
holds true with digital. You don't need more pixels, you need a larger
sensor, giving you larger pixels.

Well, you need both.

It turns out that since lenses scale up in resolution as you scale them down
in size, the improvements from larger formats in digital are much more
subtle than the improvements from smaller film. The tiny dcams actually do
fly at resolutions up to at least 5 or 6MP.

Unfortunately, there aren't many choices to give you significantly larger
sensors without going to an SLR. And of course, with larger sensors,
depth of field becomes more of a problem, but you're up against the laws
of physics here.

Actually, it turns out that _maximum_ DOF (the DOF at the point you don't
want to stop down any more because diffraction will degrade your image at
any smaller f stop) is exactly the same _whatever the size of the sensor_ as
long as you are comparing cameras with the same number of pixels.

People with more sense than money who own cameras like the 1Dsmk2 or 5D will
see a lower max DOF than users of 6 or 8MP smaller cameras, but only because
they have higher resolution requirements (in lines per picture height terms)
than the smaller sensor cameras.

David J. Littleboy
Tokyo, Japan


.

Relevant Pages

  • Re: A Step Backward ?
    ... per pixel is exactly the same in both cameras at the diffraction limit ... so the DSLR superiority is such that it allows them to match ... pixels that you so frequently rail against. ... one which bumps into or exceeds the diffraction limit you mentioned. ...
    (rec.photo.digital)
  • Re: Full-frame or 1.5 DSLR?
    ... >16.7) FF cameras can resolve subject detail as well as the 20D (or ... >But the lens has to deliver its resolution to the sensor, ... >matter that the sensor has more pixels per mm in the 20D sensor, ...
    (rec.photo.digital.slr-systems)
  • Re: Moving on from Fuji A310 ,,,
    ... > effective pixels. ... Fuji used an unusually designed CCD in some of its older cameras ... since the interpolation created twice as many "pseudo" pixels as ... actually existed on the camera's sensor. ...
    (rec.photo.digital)
  • Re: Resolution
    ... > Minolta F300 on the Imaging Resource Comparometer ... > with other cameras. ... > the sensor size increase with increased pixels. ...
    (rec.photo.digital)
  • Re: Moving on from Fuji A310 ,,,
    ... >> pixels, but also 5.2 m effective pixels. ... > since the interpolation created twice as many "pseudo" pixels as ... > actually existed on the camera's sensor. ... With these cameras, it was ...
    (rec.photo.digital)