On 5/17/2010 10:02 AM, Igor Roshchin wrote:
Larry, as far as the glass is concerned, -
there are more parameters/constrains than just lpi:
I'm aware of that, I even know how to spell MTF.
the maximum aperture, the minimum aperture (which becomes important
for the small sensor sizes, - as you end up getting diffraction),
performance at different apertures, etc.
exactly.
More over, the parameter space that determines lpi is multi-dimentional.
Those other parameters (e.g. the minimum aperture), - also can affect
lpi.
Things become even more complicated for zoom lenses, where the
parameter space blows up by probably an order of magnitude.
But I'm talking in general about a system, nominally one with
interchangable lenses.
On the side of the camera, - it is also not just the resolution of the
sensor. (E.g. high ISO performance can be separate from the resolution
optimization.)
True.
So, optimization is not (and should not be) done by just using lpi
vs. cost alone.
Exactly. Back in the silver age, if you wanted more resolution in your
photos, you pretty much had to use bigger film, though you could get a
little bit more with slower film. That's why folks used MF and view
cameras.
Today, it's not so simple. Pixel densities on the sensors probably
aren't quite down to the wavelength of light, but I've read arguments
that a lot of them actually exceed the theoretical maximum possible for
a lens to resolve.
The high level design specification would be something like:
Design an interchangeable lens digital camera system that provides the
highest possible resolution for the money in five years, taking into
account the costs of lens design, assuming that sensor costs will
continue dropping at the current rate. What is the optimal sensor size
to design for?
Igor
Mon May 17 11:37:04 CDT 2010
Larry Colen wrote:
I have the bad habit of thinking about the design of camera equipment
when I'm driving. For a mass market system, one would expect that the
engineers would figure out how much resolution is needed, and design the
system around that. Another option is to place the constraint on the
size of the sensor, and try to optimize there.
If we can assume that Moore's law is going to hold on sensors, for at
least a few more iterations, then for long term planning, sensor cost is
not going to be an issue, and it might be best to optimize system design
on the glass. Lens quality has improved over the years, but I'm
guessing that the improvement over time is more linear, or logarithmic
rather than the exponential improvement (2dB/year) of Moore's law.
The sharper a lens (lines per inch) the more expensive it is to produce.
Likewise the larger the sharp area of a lens, (24x36 sensor vs APS) the
more expensive it is to produce. There are also physical limitations to
the resolution, it can't get any better than the wavelength of light.
So, for ultimate system resolution, is there a sweet spot for lens
design where you get the best lines per inch times sensor area for the
least cost? Or is it simply a case that as far as glass is concerned,
it's always cheaper to get the same total resolution with a bigger
target area?
--
PDML Pentax-Discuss Mail List
[email protected]
http://pdml.net/mailman/listinfo/pdml_pdml.net
to UNSUBSCRIBE from the PDML, please visit the link directly above and follow
the directions.
