Phil Cook said:Quoted message said:I don't pretend to understand what on earth he's on about. But it
should cause the techies some food for thought!
I know what he's on about, and I still don't agree. It's partly to do
with colour interpolation, which I'll deal with first. There's a myth
that cameras which use Bayer to interpolate colours (most digital
cameras) only have 1/3 to 1/4 of the resolution that they claim to have,
thus a 6MP camera is really only 1.5-2MP. This is complete bolleaux.
It's only the colour that's interpolated, not the luminosity, and it's
the luminosity that creates the sharpness, not the colour.
Quoted message said:The snipped portion doesen't say a great deal except bafflegab.
Some marketing bumph makes a big deal out of colour resolution, for
example the Sigma cameras which claim to be 10MP (because they have
three colour layers in the sensor) but which are really only 3.3MP.
This Nikon bumph is basically doing a similar thing, they've worked out
a better method for resolving colour so are claiming a better
resolution, which it doesn't actually have. The colour quality may
indeed be better, but the overall image can't be any higher in
resolution.
The reality is that the human eye can't distinguish colour resolution
(sharpness) very well at all, so the "low-res" bayer colour combined
with "high-res" luminosity data actually works very well.
This is easy to test. Load an image into PS (ideally a film scan,
rather than a bayer image), switch to lab mode, then look at each
channel separately. The L (lightness/luminosity) channel is a black and
white image, and is just as sharp as the full colour image. The a and b
channels contain the colour information. Looked at own their own it's
almost impossible to see what the image is. Colour alone means very
little to the human eye.
Now select each colour channel in turn and blur it slightly. Then look
at the whole image once more, it still looks just as sharp! The "colour
quality" may not be so good, but the sharpness, i.e. the resolution, is
not affected. Our eyes don't judge sharpness by colour, that's why
black and white photogaphy is perfectly acceptable to our eyes.
Go one step further, take each colour channel and blur them beyond any
recognition, then look at the whole image. The colour will look very
strange, but with no loss of sharpness! So the idea that interpolated
colour reduces resolution is a myth, and consequently, any method of
improving colour quality in a digital sensor can not increase the
resolution!
Quoted message said:The
only actual reference to tests is this:
Quoted message said:I can't corroborate that statement but I
have tested the D2H against other Nikon Digital SLRs and the results
showed finer image detail per pixel and superior colour saturation.
Finer image detail per pixel? Nonsense. How can you have more detail
in a pixel? A pixel is a single colour, you can't have more detail in
it. Of course, that's probably not what he meant, he probably meant a
sharper image overall. This would be due to the pixel switching he
mentioned, only activating one column of pixels at a time, reducing
"bleed" to adjacent pixels. This may indeed make the image sharper
compared to previous Nikon sensors, but that's my point, he was
comparing it to other Nikon sensors. So all he's really saying is that
this is a better Nikon sensor, not that it's necessarily better than
other sensors, such as Canon's renowned Digic sensor.
Quoted message said:All the rest is manufacturer's hype and hearsay.
Yep.
The big problem here seems to be the definition of "resolution". In the
film world, resolution tends to refer to lens sharpness, whereas in the
digital world, people use the word resolution to refer to the number of
pixels in an image (we all know about monitor resolutions, don't we?).
The article that Andy quoted intentionally takes advantage of this
confusion, when he says; "I'll deal first with the resolution. A myth
about digital photography is that megapixels are everything." Now while
I'll be the first to agree that MP are not "everything", they are the
*limiting factor* where resolution is concerned.
Colour quality is a separate issue to resolution, as is noise and
dynamic range. This sensor produces better quality colour and lower
noise, and perhaps also sharpness (compared to other Nikon sensors), but
none of those increase the resolution. The resolution (in the digital
sense) is 4MP. It can never be more than 4MP, no matter how good the
sensor.
As for the "sharpness" within the image (by which I mean the other kind
of resolution), that's mainly determined by the lens quality, and to
some extent the sensor quality, and also other factors such as inbuilt
noise reduction and in-camera sharpening. It's true that one 4MP camera
may produce sharper images than another 4MP camera, but that doesn't
mean that the one camera has a better resolution than 4MP, it just means
that the other camera produces less sharp images. A 4MP camera can't
have resolution "equivalent to a 6MP camera", it's physically
impossible. But a 6MP camera may not be as sharp as it could be,
depending on lens quality etc.
In order to avoid confusion on this issue I think it's best to use the
word "resolution" to mean what most people think it means, i.e. the
number of pixels in an image, and the word "sharpness" to refer to the
variable degrees of definition within an image, which can be affected by
many different factors.
All cameras tend to use some kind of in-camera sharpening when producing
jpegs, and given most people's apparent tendency to avoid RAW
processing, I suspect that many photographers may be basing their
impression of sharpness on pre-sharpened jpegs rather than unsharpened
RAW files. In fact I recall that a comparison between the 300D and the
D70 on dpreview raved about the "great sharpening algorithm" in the
Nikon, and showed a series of comparative shots, all taken from
pre-sharpened jpegs, concluding that the D70 was sharper than the 300D!
It wouldn't surprise me if photographers are making similar comparisons
with the D2H.
Paul
--
http://www.wilderness-wales.co.uk
http://www.wildwales.fsnet.co.uk
