I'm really sorry if you all eel 'lectured to' but there is a fundamental issue that you are not grasping, which is that cameras are for making viewable images, not pixel peeping. This is at the centre of this discussion (I've re-ordered the discussion below, to keep the thread of discussion)...
Diffraction effects are related to pixel spacing and f ratio. The
diffraction disk grows larger as the lens is stopped down, and so to
avoid having the disk cover several pixels one would not want to stop
a lens down more than
This is at th nub of this issue - why do we not want the disk to cover several pixels? What is important is how sharp is the end image, regardless of how many pixels are covered. A fine pitch sensor may render the disk using a lot of pixels, but the final, viewed image of that disk will still be better than the same disk rendered with fewer pixels.
For the D3: pixel spacing = 8.5 microns
For green llght near the center of the visible spectrum: wavelength =
.55 microns.
The critical f stop for the D3 is, therefore, about f/13.
...and once youve grasped this you ralise that this is not a 'critical f stop at all', the critical f stop is set by how much resolution you want in the final viewed image, not by the sensor geometry.
One can argue that a diffraction disk that covers an area 2 pixels
wide (i.e. 4 pixels in area) is acceptable - but in that case why not
go for fewer pixels with larger area and reap the benefits of greater
DR and less noise? Clearly diffraction effects are real and to be
avoided if practical. One might need greater DOF, or to compensate
for very bright light, but one cannot say that diffraction can be
ignored unless image sharpness is also to be ignored.
...and I've covered that above.
For a given number of mega pixels and sensor technology, an FX sensor
has clear advantages over a DX sensor if the same lens is used on
both cameras and the lens in question has sufficient FOV to
illuminate the FX sensor. The advantages are in DR, apparent
sharpness, and range of usable f ratios.
Sorry this is incorrect, theoretically and practically. The difference is that the DX sensor captures a smaller part of the image projected by the lens. In terms of DR, sharpness and range of usable f ratios, they are the same. If you magnify the images to an equal size print of image, then the FX sensor will produce a better image than the DX one, because the maginification factor is less, but it will be a different image.
Even if we want to compare two lenses of the same optical quality,
but chosen for the same scene area (say 75mm of an FX camera, and
50mm on a DX camera), the advantage is on the side of the FX camera
if you have the same number of pixels total on both sensors.
Yes, now it is, because the FX sensor is collecting more light to image the scene. The number of pixels is immaterial, so long as it sofficient to render the scene with greater than the presentation resolution.
Yes, I do realize that I have not expounded on larger sensors WITH
more pixels, nor on using lenses wide open- but the rudiments of why
you might want a larger sensor with larger pixel sites are given
above.
Yes, I'll return to these now.
Sensor size, pixel spacing, and number of pixels are all related. It
appears that repeated references to one of these factors being
changed, without saying which other one is being held constant has
generated endless confusion, complications, and room (sometimes) for
questionable conclusions.
Yes, because they are not tightly related. A sensor designer has (apart from semiconductor process issues) two variables to play with - sensor size and pixel pitch. The fact that sensor size determines the number of pixels of a given pitch that can be packed into it is not to say that all properties of the sensor are to do with pixel pitch. Many are to do with size alone. Some are to do with pixel pitch. The various contributors to this thread have been playng fast and loose with the distinction and claiming that its Ok because both are 'related'. They are only related if one chooses to constrain the number of pixels on a sensor to a particular value.
For a given number of pixels,
There you go now - what's a 'given number of pixels' got to do with anything?
a larger sensor will have more distance
...
DR and better low light performance.
And now you've made the mistake from your conflation of different issues. DR and low light performance is a function of the light gathered by the sensor, not per pixel. Apart from the issue of fill factor, which perhaps favours larger pixels but is marginal at DSLR geometries, equal size sensors using equivalent technologies will yield the same DR and low light performance (at an image level) regardless of the number of pixels.
These results do not follow from sensor size per se: a larger sensor with more pixels and the
same pixel size would not, for a given sensor type, have more DR or
sensitivity - but it would have more IQ and more pixels.
Sorry, you are wrong about this at any other than the pixel peeping level. For equal size areas on the final image, both sensors collect the same number of photons, therefore, so long as the technology is equivalent, both will produce the same DR and have the same sensitivity.
In the context of D3 vs D300, the D3's design parameters mean less
demand on lens resolution, greater range of f stops without
diffraction, greater DR, and less noise.
All of these are factors of sensor size, and the consequent use of lower levels of magnification to view the image. If you maintain otherwise, please let me know the mechanism by which these improvements occur - I have given the reasons behind my theory, you have not done the same for yours.
Best wishes to all
-George
--
Bob
