So how good is 36 megapixels?

Alan Browne wrote:
On 2013.03.20 19:17 , Doug McDonald wrote:
On 3/20/2013 4:33 PM, Alan Browne wrote: On 2013.03.18 20:09 , Doug
McDonald wrote:

I posted this because I was truly amazed at how bad it was. For normal
scenes its a non-issue.

Recall that you put up these images to refute the idea that lenses at
some point can't deliver detail to the ever increasing density of
sensors.

In reply to me stating that lenses behaved as spatial f filters as
sensor densities got higher,

You said: "Absurd: I own an 18 mpixel Canon7D, a crop frame
camera: and it shows bad moire on subjects like an LCD
TV. And it has an AA filter."

Your photos only illustrate a grid interference pattern. Nothing to to
at all with decreased resolution due to a lens and high sensor
densities. Nothing to do with AA filters either, to put a point on it.


You are wrong. Moire IS a "grid interference pattern". The frequency of

No **** Sherlock.

You're completely ignoring the point about lens _MTF_
which is what I referred to before your "absurd"
statement.

His point is well taken, and does not ingore MTF as opposed
to showing that the MTF is indeed high enough to be ignored.

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

On 3/20/2013 6:34 PM, Alan Browne wrote:
The pictures show CONCLUSIVELY that the lens is fully adequate to
resolve adjacent pixels, and that the AA filer is NOT seriously smearing
out the image enough to stop moire.

If the AA smeared it enough to remove the moiré that you captured in
your contrivation, then you would be just as happy with the detail of a
6 Mpix camera. Probably not even then.



I never said that I wanted no moire in this absolutely
utterly
worst case moire test. What makes you think I did? I LIKE the camera
the way it is. I just pointed out that if one tried to get it
to generate moire, it will do so.

Of course a lens with a crappy MTF will reduce moire. But my
trusty 24-105 f/4L zoom, not exactly the best L glass ever,
still generates some moire.

Doug McDonald

"Wolfgang Weisselberg" wrote in message
...

Trevor wrote:

No, it also depends on how much improvement can be made in lens design,
or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

For quality lenses used maximizing their capability? Where is
that XXX-Mpix or XXXX-MPix sensor you're talking about used
in commercial FF DSLRs?

Yes. The D800 is 100 pixels/mm, which is 50 lp/mm, and film types with
high-contrast targets (locked down on a granite optical bench, Tech Pan, big
strobes) claim to see well over that all the time.

http://www.hevanet.com/cperez/MF_testing.html

I'm pretty sure that _for some uses_, 36MP is a quite sensible thing in an
FF camera, and a lot more than merely a marginal improvement over 20MP.
Landscapes where you don't need all that much DoF and are making 30x40"
prints that people will walk up to. I see a lot of posters that size in the
Tokyo train/subway stations that are clearly done in large-format film;
amazing quality. Maybe the D800 will compete with that. But at f/8 and with
very heavy tripods.

--
David "My tripod is heavier than yours" Littleboy
Tokyo, Japan

"David J. Littleboy" wrote:
"Wolfgang Weisselberg" wrote in message
...

Trevor wrote:

No, it also depends on how much improvement can be
made in lens design, or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

For quality lenses used maximizing their capability? Where is
that XXX-Mpix or XXXX-MPix sensor you're talking about used
in commercial FF DSLRs?

Yes. The D800 is 100 pixels/mm, which is 50 lp/mm, and

The D800 has 7360 pixels across a 35.9mm sensor, which
is 205.0 pixels per mm, and 102.5 line pairs per mm.

film types with high-contrast targets (locked down on a
granite optical bench, Tech Pan, big strobes) claim to
see well over that all the time.

http://www.hevanet.com/cperez/MF_testing.html

I'm pretty sure that _for some uses_, 36MP is a quite
sensible thing in an FF camera, and a lot more than
merely a marginal improvement over 20MP. Landscapes
where you don't need all that much DoF and are making
30x40" prints that people will walk up to. I see a lot
of posters that size in the Tokyo train/subway stations
that are clearly done in large-format film; amazing
quality. Maybe the D800 will compete with that. But at
f/8 and with very heavy tripods.

Exactly! The 36MP image when printed at 300 PPI
produces roughly a 16x24 print. Hence anyone commonly
making 16x20's or larger will appreciate the D800. A
24.7MP D600 image printed at 300 PPI is about 14x20 and
will be almost as good up to 16x20, but not at larger
sizes. A D4 with 16.2MP prints at about 11x16, and
hence starts losing it at anything larger than a 11x14.

Obviously it is often possible to print with fewer than
300 PPI, so these are ballpark sizes, but by the same
token all of them are without any cropping (other than
to the state aspect ratio) either.

To put that into perspective, there absolutely will be
a market for a fullframe DSLR that produces 50MP or more.

If the D7100 24MP sensor were, as an example, scaled up
to fullframe it might have about 9167x6111 (56MP) pixels,
and would make, at 300 PPI, a 20x30 print. A $6000 DSLR
like that would sell like hotcakes. A $3000 DSLR would
make the popularity of the D800 fade forever!

--
Floyd L. Davidson http://www.apaflo.com/
Ukpeagvik (Barrow, Alaska)

On 22/03/2013 4:33 a.m., Floyd L. Davidson wrote:
"David J. Littleboy" wrote:
"Wolfgang Weisselberg" wrote in message
...

Trevor wrote:

No, it also depends on how much improvement can be
made in lens design, or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

For quality lenses used maximizing their capability? Where is
that XXX-Mpix or XXXX-MPix sensor you're talking about used
in commercial FF DSLRs?

Yes. The D800 is 100 pixels/mm, which is 50 lp/mm, and

The D800 has 7360 pixels across a 35.9mm sensor, which
is 205.0 pixels per mm, and 102.5 line pairs per mm.

film types with high-contrast targets (locked down on a
granite optical bench, Tech Pan, big strobes) claim to
see well over that all the time.

http://www.hevanet.com/cperez/MF_testing.html

I'm pretty sure that _for some uses_, 36MP is a quite
sensible thing in an FF camera, and a lot more than
merely a marginal improvement over 20MP. Landscapes
where you don't need all that much DoF and are making
30x40" prints that people will walk up to. I see a lot
of posters that size in the Tokyo train/subway stations
that are clearly done in large-format film; amazing
quality. Maybe the D800 will compete with that. But at
f/8 and with very heavy tripods.

Exactly! The 36MP image when printed at 300 PPI
produces roughly a 16x24 print. Hence anyone commonly
making 16x20's or larger will appreciate the D800. A
24.7MP D600 image printed at 300 PPI is about 14x20 and
will be almost as good up to 16x20, but not at larger
sizes. A D4 with 16.2MP prints at about 11x16, and
hence starts losing it at anything larger than a 11x14.

Obviously it is often possible to print with fewer than
300 PPI, so these are ballpark sizes, but by the same
token all of them are without any cropping (other than
to the state aspect ratio) either.

To put that into perspective, there absolutely will be
a market for a fullframe DSLR that produces 50MP or more.

If the D7100 24MP sensor were, as an example, scaled up
to fullframe it might have about 9167x6111 (56MP) pixels,
and would make, at 300 PPI, a 20x30 print. A $6000 DSLR
like that would sell like hotcakes. A $3000 DSLR would
make the popularity of the D800 fade forever!

But...
OK - here's the DXOMark article:
http://www.dxomark.com/index.php/Pub...l-can-help-you

"PPI = 300" isn't a measure or standard of print quality (detail,
sharpness etc).

Wolfgang mentions above that there's a "41mp" phone camera producing
7152 x 5368 pixel images. Of course "normally" this is scaled back to
3, 5, or 8 megapixels - but ~ 38mp is the native sensor output.
(They've probably been told millions of times not to exaggerate).
So then you can make 24 x 18 prints from the Nokia 808 cellphone, which
would just have as many or more PPI as a D800 print that size, or as the
generally accepted rule is that 300dpi is about right, then at that size
the Nokia 808 print would be practically indistinguishable from a medium
format digital image from an H5D, scanned large format film, or a
trillion gigapixel stitched mosaic... I don't think so.
DXO don't seem to have a P-megapixel figure for the 808. It would be a
little bit difficult to define in this case, as with digital zoom with a
varying rate of oversampling to a downscaled image of standard size in
normal use, it will be all over the place depending on "focal length".
The (crops of) 38mp ex-sensor images I've looked at are course soft,
noisy mush, but as a 5mp or so "P-megapixel" camera, it seems okay.

So, "PPI" is a waste of time for use in determining maximum print size -
as there's something missing. That something is "Perceptual" (mega)pixels.

"PPPI" for sensor/lens combinations would be a much better guide to
maximum print size at optimum print quality using a particular
sensor/lens combination.

Using that, then there's almost no difference between the D3x and the
D800. You could print about 5% longer and wider at the same print
quality, or crop the image 5% more, but only of you use the very best
lens DXO tested on the D800.

Hopefully, DXO also produce "P-megapixel" results for the D800E, then we
can see if there's a benefit to the D800E missing out on an AA filter,
to offset the inevitable increase in image flaws as a result of not
having one.

The D3x/D800 comparison seems to show that 50 or 100mp Fx is mainly a
waste of time (perceptual resolution), but on the other hand, if not
having an AA filter becomes the norm, then the higher the pixel density,
then the better.

On 2013.03.20 22:16 , Doug McDonald wrote:
On 3/20/2013 6:34 PM, Alan Browne wrote:
The pictures show CONCLUSIVELY that the lens is fully adequate to
resolve adjacent pixels, and that the AA filer is NOT seriously
smearing
out the image enough to stop moire.

If the AA smeared it enough to remove the moiré that you captured in
your contrivation, then you would be just as happy with the detail of a
6 Mpix camera. Probably not even then.



I never said that I wanted no moire in this absolutely
utterly
worst case moire test. What makes you think I did? I LIKE the camera
the way it is. I just pointed out that if one tried to get it
to generate moire, it will do so.

Your "absurd" declaration did not, in any way, disprove what I said
about lens MTF softening as desired sampling (sensor density) goes up.

Further, what I said above wrt to AA filters was not meant to be what
you desire, just to illustrate the sort of heavy AA filter that would be
required to tame the moiré in your high school lab experiment.

(Another way of showing the irrelevancy of your examples if I have to
make that plain to you.)

Of course a lens with a crappy MTF will reduce moire. But my
trusty 24-105 f/4L zoom, not exactly the best L glass ever,
still generates some moire.

In summary, you retract what you said about my statement of lens mtf
acting as a softening filter where ever increasing sensor densities are
occurring.

Thought so. Next time just keep your hands off the keyboard and save
everyone some bandwidth.

--
"There were, unfortunately, no great principles on which parties
were divided – politics became a mere struggle for office."
-Sir John A. Macdonald

"Floyd L. Davidson" wrote in message
...
"David J. Littleboy" wrote:
"Wolfgang Weisselberg" wrote in message
...
Trevor wrote:

No, it also depends on how much improvement can be
made in lens design, or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

Right, and for most available lenses once you get to 50mp and beyond in 35mm
format. Which was my point.
We should see a return in development and sales toward prime lenses, as
zooms certainly won't cut it, and many of the primes will need improving to
do so as well.


For quality lenses used maximizing their capability? Where is
that XXX-Mpix or XXXX-MPix sensor you're talking about used
in commercial FF DSLRs?

Yes. The D800 is 100 pixels/mm, which is 50 lp/mm, and

The D800 has 7360 pixels across a 35.9mm sensor, which
is 205.0 pixels per mm, and 102.5 line pairs per mm.

film types with high-contrast targets (locked down on a
granite optical bench, Tech Pan, big strobes) claim to
see well over that all the time.

http://www.hevanet.com/cperez/MF_testing.html

I'm pretty sure that _for some uses_, 36MP is a quite
sensible thing in an FF camera, and a lot more than
merely a marginal improvement over 20MP. Landscapes
where you don't need all that much DoF and are making
30x40" prints that people will walk up to. I see a lot
of posters that size in the Tokyo train/subway stations
that are clearly done in large-format film; amazing
quality. Maybe the D800 will compete with that. But at
f/8 and with very heavy tripods.

Exactly! The 36MP image when printed at 300 PPI
produces roughly a 16x24 print. Hence anyone commonly
making 16x20's or larger will appreciate the D800. A
24.7MP D600 image printed at 300 PPI is about 14x20 and
will be almost as good up to 16x20, but not at larger
sizes. A D4 with 16.2MP prints at about 11x16, and
hence starts losing it at anything larger than a 11x14.

Obviously it is often possible to print with fewer than
300 PPI, so these are ballpark sizes, but by the same
token all of them are without any cropping (other than
to the state aspect ratio) either.

To put that into perspective, there absolutely will be
a market for a fullframe DSLR that produces 50MP or more.


Right, but only with lenses that are capable of providing resolution between
those pixels, otherwise you could just as easily use a 24-36mp camera and
upscale in PS for the same 300PPI you deem necessary. Either way you don't
get a "real" 300PPI *resolution* image if the lens can't manage it.



If the D7100 24MP sensor were, as an example, scaled up
to fullframe it might have about 9167x6111 (56MP) pixels,
and would make, at 300 PPI, a 20x30 print. A $6000 DSLR
like that would sell like hotcakes. A $3000 DSLR would
make the popularity of the D800 fade forever!

Absolutely, and most would not care if their lenses were not as good as the
camera, the boasting rights alone would sell many. Which is not to say that
it would not still be great for use with those few lenses which would allow
it to approach full resolution of course!

Trevor.

David J. Littleboy wrote:
"Wolfgang Weisselberg" wrote in message
Trevor wrote:

No, it also depends on how much improvement can be made in lens design,
or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

For quality lenses used maximizing their capability? Where is
that XXX-Mpix or XXXX-MPix sensor you're talking about used
in commercial FF DSLRs?

Yes. The D800 is 100 pixels/mm, which is 50 lp/mm, and film types with
high-contrast targets (locked down on a granite optical bench, Tech Pan, big
strobes) claim to see well over that all the time.

http://www.hevanet.com/cperez/MF_testing.html


In addition, [scanning film]:

| The tests here were at f/11 with the 35mm, giving over 18,000 dpi at
| the Dawes limit, or 17000 x 26,000 pixels! This is more than the 4x5
| test done here. The 4x5, at f/45 gives a Dawes limit of 4516 DPI,
| more than the test here, but the 3300 DPI 4x5 scan is close to the
| Rayleigh limit. This shows that the film in both the 35mm and the
| 4x5 is the limit, not the lens. If much finer grained film could be
| made, the lens system could deliver the image detail and 35mm would be
| comparable to existing 4x5, or over 600 megapixel equivalent!
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
http://www.clarkvision.com/articles/...ail/index.html

So here's 600 MPix scanned (i.e. full RGB triples), which
is at least 848 MPix Bayer-pattern for identical luminance
resolution --- and more for identical chrominance resolution.

-Wolfgang

Trevor wrote:
"Floyd L. Davidson" wrote in message
"David J. Littleboy" wrote:
"Wolfgang Weisselberg" wrote in message
Trevor wrote:

No, it also depends on how much improvement can be
made in lens design, or
it's just a number for the sake of a number. Sensor design is already
outstripping lens capability,

For mobile phone 41 MPix sensors, yes.
For the tele end of superzoom 16 M Pix cameras, yes.
For some kit lenses, especially outside their best
performance, yes.

Right, and for most available lenses once you get to 50mp and beyond in 35mm
format. Which was my point.

Which is your claim. To have a point, you'd have to prove
your claim first.

BTW, most available lenses are mobile phone lenses, followed
by compact camera lenses, so, well, yes, you might have a
point there, just not the point you wanted to make.

We should see a return in development and sales toward prime lenses, as
zooms certainly won't cut it,

Please find out which are the best wide angle lenses in FF
and crop cameras. You'll be surprised.

and many of the primes will need improving to
do so as well.

See my other post --- the one about scanning film.

Exactly! The 36MP image when printed at 300 PPI
produces roughly a 16x24 print. Hence anyone commonly
making 16x20's or larger will appreciate the D800.

BZZZT. Not "anyone".
You have to work painstakingly to appreciate the D800 (not
everyone is willing or able to do so) --- and other people are
using large format for a reason and therefore won't appreciate
the very low resolution of the D800.


To put that into perspective, there absolutely will be
a market for a fullframe DSLR that produces 50MP or more.

Of course. It's those that shoot medium format now because
they need the resolution, want more mobility and don't need
stuff the FF-DSLR can't offer.

Right, but only with lenses that are capable of providing resolution between
those pixels,

The fact that you don't see much difference in the centre
between average-to-good and exceptional lenses at f/8 shows
that not even good lenses are pressed hard there, never mind
exceptional ones.

If the D7100 24MP sensor were, as an example, scaled up
to fullframe it might have about 9167x6111 (56MP) pixels,
and would make, at 300 PPI, a 20x30 print. A $6000 DSLR
like that would sell like hotcakes. A $3000 DSLR would
make the popularity of the D800 fade forever!

Absolutely, and most would not care if their lenses were not as good as the
camera, the boasting rights alone would sell many.

Any lens-camera combination improves when increasing the
resolution of the sensor. E.g. compare the lp/ph for the
D800 and another FF camera with less pixels for the same
lens.

-Wolfgang

Me wrote:

But...
OK - here's the DXOMark article:
http://www.dxomark.com/index.php/Pub...l-can-help-you

"PPI = 300" isn't a measure or standard of print quality (detail,
sharpness etc).

No, it's an upper limit to resolution.

That's why there's something like the SQF (subjective quality
factor).

Wolfgang mentions above that there's a "41mp" phone camera producing
7152 x 5368 pixel images.

In fact, it doesn't, from what I hear --- it only produces 8
MPix, as it's --- as I understand it --- scaled down before
the result is handed to the user.

Of course "normally" this is scaled back to
3, 5, or 8 megapixels - but ~ 38mp is the native sensor output.
(They've probably been told millions of times not to exaggerate).

So then you can make 24 x 18 prints from the Nokia 808 cellphone, which
would just have as many or more PPI as a D800 print that size,

If you could get the raw sensor output before it's scaled
down, that is, and if you didn't care too much about the
quality of each teeny-weeny pixel (never mind the lens
limitations).

or as the
generally accepted rule is that 300dpi is about right,

It's only right when the PPI are really carrying information
and the viewing distance doesn't get too short.

And for images where people get really close, you want more
than just 300 PPI.

then at that size
the Nokia 808 print would be practically indistinguishable from a medium
format digital image from an H5D, scanned large format film, or a
trillion gigapixel stitched mosaic... I don't think so.

http://www.luminous-landscape.com/reviews/kidding.shtml

So, "PPI" is a waste of time for use in determining maximum print size -
as there's something missing. That something is "Perceptual" (mega)pixels.

Perceptual megapixels is bull. There is --- as you said --- no
eqivalence between megapixels and print quality. Especially
where print size and print quality don't easily relate.
Is the difference between 280 and 300 PPI visible? Is the
difference between 140 and 150 PPI visible? What is the effect
of sharpening (increasing acutance) on the perceptual megapixels
(hint: the percepted quality is strongly affected!)? What is
the effect on different print sizes? (Hint: an aberrated f/2.8
lens at f/2.8 can be sharper at 3.5"x5" than one stopped down
to f/8 --- and the other way round at 20"x30". See
http://www.bobatkins.com/photography.../mtf/mtf4.html
)

So what you really want is (something like) SQF, which has
been tested by time and is well understood. (A difference of
5 starts to become visible; SQF can be used to precisely
answer questions like, “How much larger can I print with a
12.8 Megapixel full-frame DSLR than with an 8.3 Megapixel
APS-C DSLR?�)

And SQF takes into account that the eye is most sensitive
at 3-12 cycles/degree. Which perceptual megapixels don't
recognize at all.

Yes, SQF is dependent on print size, as it describes the
viewing experience. Perceptual megapixels don't.


"PPPI" for sensor/lens combinations would be a much better guide to
maximum print size at optimum print quality using a particular
sensor/lens combination.

And SQF plotted over print size ...


Using that, then there's almost no difference between the D3x and the
D800. You could print about 5% longer and wider at the same print
quality, or crop the image 5% more, but only of you use the very best
lens DXO tested on the D800.

You need to plot over the various ISO settings, apertures
(using real lenses) and print sizes, at the very least.
Just "perceptual megapixels" doesn't say enough.

BTW, what again is the formula for "perceptual megapixels"?

The D3x/D800 comparison seems to show that 50 or 100mp Fx is mainly a
waste of time (perceptual resolution),

Which is most likely because perceptual megapixels aren't
answering the question in first place.

-Wolfgang