Need help in calculating digital camera's MP

HEMI - Powered wrote:
[]
David, we've had this discussion ad nauseum, namely is "more mega
pixels better images" or not. You are obviously correct that few
cameras other than toys are less than about 6 MP but that hardly
means they are all created equal.

Indeed - how annoying might the small amounts of pincushion or barrel
distortion in many of today's zoom lenses be when viewed against the dead
straight edges on a LCD or Plasma display, if left uncorrected. I suspect
that a 1920 x 1080 HD TV display worthy of the name would provide quite a
critical environment for viewing images, if for no other reason than its
physical size. You should certainly see the benefits of a higher-quality
source image.

Cheers,
David

HEMI - Powered wrote:
[]
Huh?! How does the way a camera "sees" a pixel translate into how a
TV sees one? Or, a PC monitor? 16 million color requires 3 bytes per
pixel, of course, but how does this matter to a TV? Isn't it far more
important how the TV system that reads digital images from, say, it's
memory card slot, depend on how well it fills the screen from
whatever it is provided?

Inded, yes. What I was saying was that in the camera, what we call 2Mpix
is not 2Mpix of red, 2Mpix of green, and 2Mpix of blue, so the overall
image resolution in RGB pixels is somewhat less than 2Mpix, perhaps 1Mpix
for the sake of argument.

However, the display does have as many RGB pixels as specified (i.e. 1920
x 1080), and so it can actually display 2Mpix of RGB. So if you want the
display to be the limiting factor, send it rather more than 2Mpix of RGB,
i.e. something like 4Mpix derived from a Bayer-sensor camera.

I therefore revise my estimate to, for a 4:3 aspect ratio camera
2 * 1920 * 1440, i.e. about 5-6Mpix, resampled and cropped to a
16:9 1920 x 1080 ready-to-display image.

True enough but my Canon DSLR creates 3:2 images. But, whether a
camera is 4:3 or 3:2 or something else, to get all the way to a
16:9/16:10 aspect ratio also requires the photographer to be VERY
aware of how they must crop their images or something important will
likely be lost.

Yes, and having a 16:9 crop frame visible in the finder would be a help.
You see this in reverse in TV work, where although the cameras are 16:9,
they have the older 4:3 aspect ratio marked (perhaps even with a slight
margin), so that the "important" action isn't lost on viewers with older
sets....

Cheers,
David

On Sat, 03 Jan 2009 08:14:47 GMT, "David J Taylor"
wrote in
:

John Navas wrote:
On Fri, 02 Jan 2009 20:04:19 GMT, "David J Taylor"
[]
The LCD TVs I've seen are quite capable of pixel-level resolution,
John.

Brand and model please.

All I have tested, when fed from the computer input.

Which isn't what I was talking about.

--
Best regards,
John
Panasonic DMC-FZ8, DMC-FZ20, and several others

John Navas wrote in
:

There's another reason for using more pixels. In the display, each of
the pixels is an RGB triple, i.e. a full colour pixel, whereas in the
camera each pixel is either red, green or blue. So to match the
display resolution, you may need more pixels in the (Bayer) camera
than are on the display.

Most current TV electronics aren't capable of that kind of pixel level
resolution -- put the HD input on pause and look closely -- and your
eyes couldn't see it in a moving image even if.

We're not talking about broadcast quality; we're talking about putting
JPEG images on the HD TV's screen, in which case it behaves like a
desktop computer monitor of 1920*1080 resolution. And if you don't think
broadcast HD can be close to that, you haven't used Verizon FiOS.
Verizon does not compress their broadcasts much further (if at all) than
the source does, and freeze-frames in HD with the right source can look
pretty much like a high-Q JPEG.

I don't know if any tests have been done to check what the
actual ratio is, but I recall that the Foveon sensor is reckoned by
some to be equivalent to something like twice the number of pixels.

That's not directly comparable.

Never directly, because there is no direct match except with hypothetical
infinite Bayer resolution, but for all practical purposes, 4x the MP
pretty much does most of (the good) that Foveon does, without the Foveon
artifacts, and Sigma aliasing.

On Sat, 03 Jan 2009 15:06:00 GMT, John Sheehy wrote in
:

John Navas wrote in
:

Most current TV electronics aren't capable of that kind of pixel level
resolution -- put the HD input on pause and look closely -- and your
eyes couldn't see it in a moving image even if.

We're not talking about broadcast quality; we're talking about putting
JPEG images on the HD TV's screen, in which case it behaves like a
desktop computer monitor of 1920*1080 resolution.

We're actually talking about input from a DVD player, which is roughly
limited to 640x480.

--
Best regards,
John
Panasonic DMC-FZ8, DMC-FZ20, and several others

On Sat, 03 Jan 2009 08:25:41 GMT, "David J Taylor"
wrote in
:

nospam wrote:

the foveon fans come up with their own math to justify the sensor's
existence. it's often hilarious. i've seen everything from 1.4x to
over 3x, with some even claiming infinite resolution, depending on how
creative the (misinformed) zealot is. the main difference is not the
co-located layers, but rather the false detail from the lack of an
anti-alias filter and heavy sharpening in the raw processing.

Yes, it was the only closely related comparison I could think of at the
time, and has unfortunate connotations for the photographic community!
Perhaps the 3-CCD video camera would have been a better comparison, but
there you have alignment issues between R, G & B which are not issues for
a LCD screen. I'll accept a factor of two until I see a more reasoned
(and perhaps empirically-tested) value.

Suit yourself, but I think that's almost certainly on the high side.
It's not a simple resolution issue.

--
Best regards,
John
Panasonic DMC-FZ8, DMC-FZ20, and several others

On Sat, 03 Jan 2009 08:31:20 GMT, "David J Taylor"
wrote in
:

HEMI - Powered wrote:
[]
David, we've had this discussion ad nauseum, namely is "more mega
pixels better images" or not. You are obviously correct that few
cameras other than toys are less than about 6 MP but that hardly
means they are all created equal.

Indeed - how annoying might the small amounts of pincushion or barrel
distortion in many of today's zoom lenses be when viewed against the dead
straight edges on a LCD or Plasma display, if left uncorrected.

How would that be any different from the dead straight edges of prints?

--
Best regards,
John
Panasonic DMC-FZ8, DMC-FZ20, and several others

John Navas wrote:
On Sat, 03 Jan 2009 08:25:41 GMT, "David J Taylor"
wrote
in :

nospam wrote:

the foveon fans come up with their own math to justify the sensor's
existence. it's often hilarious. i've seen everything from 1.4x to
over 3x, with some even claiming infinite resolution, depending on
how creative the (misinformed) zealot is. the main difference is
not the co-located layers, but rather the false detail from the
lack of an anti-alias filter and heavy sharpening in the raw
processing.

Yes, it was the only closely related comparison I could think of at
the time, and has unfortunate connotations for the photographic
community! Perhaps the 3-CCD video camera would have been a better
comparison, but there you have alignment issues between R, G & B
which are not issues for a LCD screen. I'll accept a factor of two
until I see a more reasoned (and perhaps empirically-tested) value.

Suit yourself, but I think that's almost certainly on the high side.
It's not a simple resolution issue.

Do you know of any studies into pure RGB versus Bayer? As you say, there
will be more to it than just resolution. Just to clarify - two in pixel
count means sqrt (2) in linear resolution.

David

John Navas wrote:
On Sat, 03 Jan 2009 08:31:20 GMT, "David J Taylor"
wrote
in :

HEMI - Powered wrote:
[]
David, we've had this discussion ad nauseum, namely is "more mega
pixels better images" or not. You are obviously correct that few
cameras other than toys are less than about 6 MP but that hardly
means they are all created equal.

Indeed - how annoying might the small amounts of pincushion or barrel
distortion in many of today's zoom lenses be when viewed against the
dead straight edges on a LCD or Plasma display, if left uncorrected.

How would that be any different from the dead straight edges of
prints?

You can trim the print to match the image! But I was thinking back to the
OPs need.

David

On Sat, 03 Jan 2009 16:53:51 GMT, "David J Taylor"
wrote in
:

John Navas wrote:

Suit yourself, but I think that's almost certainly on the high side.
It's not a simple resolution issue.

Do you know of any studies into pure RGB versus Bayer? As you say, there
will be more to it than just resolution. Just to clarify - two in pixel
count means sqrt (2) in linear resolution.

Please see my other post, with comments on luminance, chrominance and
eye sensitivity.

--
Best regards,
John
Panasonic DMC-FZ8, DMC-FZ20, and several others