Nikon new release D7100

Eric Stevens wrote:
On Sun, 10 Mar 2013 14:48:42 -0700, Savageduck
wrote:

On 2013-03-10 13:56:05 -0700, Eric Stevens said:

On Sat, 9 Mar 2013 19:45:01 -0800, Savageduck
wrote:

On 2013-03-09 19:20:01 -0800, Rob said:

On 10/03/2013 12:37 PM, Floyd L. Davidson wrote:
Wolfgang Weisselberg wrote:
At a back-of-the-envelope calculation, you'd need about 0.25
nm pixel size for f/1.2; which means about 96,000 x 144,000
pixels at FF, which comes to 13824 MPix (about 14 GPix).
Somehow 38 MPix seems a bit low compared to that.

In fact 150MP is getting close, and 250MP is certainly
enough that no anti-aliasing filter will be needed.
That is based on a diffusion limited lens with an f/1.2
aperture.

Hence something between 175 and 200MP is a pretty fair
estimate for the pixel count needed to simply eliminate
the need for any anti-aliasing filter on a 36x24mm
sensor.

Which is interesting, given that it would appear that
technology is just around the corner and will likely be
available in the next decade.



The next FF nikon looking likely to be 44Mp, now that's a worry,
thinking storage space.

Not to worry. It will come with its very own SATA SSD slot/dock.
...but no CF or SDHC slot. ;-)

How many seconds per shot on Continuous High?


The 50GB buffer will handle that issue. ;-)

Well beyond the hypothetical fantasy consider the following:

If you consider that many SSD's currently have read/write specs at
559/527 MB/s, a 44MB file should be written in 0.0835 seconds. A 10
shot continuous burst should be captured in less than 1 second. It is
the buffer together with the card write speed which limits your
continuous high burst shot capacity.

You have to fill the buffer first and it is noticable that cameras
with large sensors are slower to fill the buffer than cameras with a
smaller sensor. I don't know where the bottle neck actually is but I
suspect that is the processing off the sensor.

Processing speed determines the maximum frame rate. The size of the
buffer compared to the size of each image determines how many shots
can be in a burst. The write speed to the memory card determines
how soon a consecutive burst can be started.

Given that you can currently buy a 480GB OWC SSD for $569, compared to
the fastest UDMA 32GB CF cards with write speeds of 145MB/s. at B&H
sale prices of $132.95. So even if compared to a 120GB SSD @ $150 the
equivalent capacity (let's say 4 x 32GB UDMA-7 CF cards, not an
unreasonable purchase for a pro) at a quarter the speed would cost that
pro $531.80.

...and since the CF card is nothing but a slowish, miniaturized SSD,
why not develop an appropriately dimensioned high performance SSD for
pro-camera use?

No doubt something like this will eventually happen but even now we
are pushing the limits in a number of respects.

The current XQD specification is for 500MB/s write
speeds, and XQD 2.0 specifications are 1000MB/s.

Wait ten years... when frame rates will be 24 fps,
images will be 200MP, and 500 shot bursts will be
possible. (Essentially high quality video for about 20
seconds.)

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

Eric Stevens wrote:
On Mon, 11 Mar 2013 09:00:32 +1100, Rob wrote:

On 11/03/2013 7:56 AM, Eric Stevens wrote:
On Sat, 9 Mar 2013 19:45:01 -0800, Savageduck
wrote:

On 2013-03-09 19:20:01 -0800, Rob said:

On 10/03/2013 12:37 PM, Floyd L. Davidson wrote:
Wolfgang Weisselberg wrote:
At a back-of-the-envelope calculation, you'd need about 0.25
nm pixel size for f/1.2; which means about 96,000 x 144,000
pixels at FF, which comes to 13824 MPix (about 14 GPix).
Somehow 38 MPix seems a bit low compared to that.

In fact 150MP is getting close, and 250MP is certainly
enough that no anti-aliasing filter will be needed.
That is based on a diffusion limited lens with an f/1.2
aperture.

Hence something between 175 and 200MP is a pretty fair
estimate for the pixel count needed to simply eliminate
the need for any anti-aliasing filter on a 36x24mm
sensor.

Which is interesting, given that it would appear that
technology is just around the corner and will likely be
available in the next decade.



The next FF nikon looking likely to be 44Mp, now that's a worry,
thinking storage space.

Not to worry. It will come with its very own SATA SSD slot/dock.
...but no CF or SDHC slot. ;-)

How many seconds per shot on Continuous High?



That's not as funny as it sounds, if you have ever used the D100, were
they sloooow, the D200 became usable.

I didn't mean it to be funny. It's a serious problem.

And a lot of people didn't consider the D200 to be
"usable" either.

The D4 is the first DSLR that I'm aware of where you can
hold down the shutter release button and rattle off 100
shots in RAW mode without significant gaps while the
buffer is cleared.

I've never done that for a real shoot, I used to hit the
limit with a D3S on a regular basis, and never do with a
D4. Once you get used to a D4 everything else can be a
little annoying now and then, even a D800 with all those
fabulous pixels.

Just wait for the D5 generation... more pixels than a D800
and faster than a D4!

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

On 2013-03-10 16:12:53 -0700, Savageduck said:

On 2013-03-10 15:37:01 -0700, (Floyd L. Davidson) said:

Savageduck wrote:
...and since the CF card is nothing but a slowish,
miniaturized SSD, why not develop an appropriately
dimensioned high performance SSD for pro-camera use?

It's called an XQD card.

Well, OK then!
So the future is sort of now, but so far only offered for the D4.
...and still a write speed of 145MB/s (1100x), which is basically the
same speed as the Lexar UDMA-7 at 145MB/s (1000x) at a cost difference
of $160. Why bother?

Note: $299 for a 32GB XQD, $579.99 for a 64GB XQD. Still about a
quarter the speed and well above the cost of equivalent capacity SSD
(60GB SSD @ $84 & 527MB/s write speed)
Still behind SSD's in both performance and cost/GB.

So my question remains, why not develop an appropriately dimensioned
high performance SSD for pro-camera use?

I mistyped that current Lexar XQD (1100x) writespeed, it is noted at
168MB/s. Better than most today but not significant over UDMA-7.

--
Regards,

Savageduck

On 2013-03-10 16:31:59 -0700, (Floyd L. Davidson) said:

Eric Stevens wrote:
On Mon, 11 Mar 2013 09:00:32 +1100, Rob wrote:

On 11/03/2013 7:56 AM, Eric Stevens wrote:
On Sat, 9 Mar 2013 19:45:01 -0800, Savageduck
wrote:

On 2013-03-09 19:20:01 -0800, Rob said:

On 10/03/2013 12:37 PM, Floyd L. Davidson wrote:
Wolfgang Weisselberg wrote:
At a back-of-the-envelope calculation, you'd need about 0.25
nm pixel size for f/1.2; which means about 96,000 x 144,000
pixels at FF, which comes to 13824 MPix (about 14 GPix).
Somehow 38 MPix seems a bit low compared to that.

In fact 150MP is getting close, and 250MP is certainly
enough that no anti-aliasing filter will be needed.
That is based on a diffusion limited lens with an f/1.2
aperture.

Hence something between 175 and 200MP is a pretty fair
estimate for the pixel count needed to simply eliminate
the need for any anti-aliasing filter on a 36x24mm
sensor.

Which is interesting, given that it would appear that
technology is just around the corner and will likely be
available in the next decade.



The next FF nikon looking likely to be 44Mp, now that's a worry,
thinking storage space.

Not to worry. It will come with its very own SATA SSD slot/dock.
...but no CF or SDHC slot. ;-)

How many seconds per shot on Continuous High?



That's not as funny as it sounds, if you have ever used the D100, were
they sloooow, the D200 became usable.

I didn't mean it to be funny. It's a serious problem.

And a lot of people didn't consider the D200 to be
"usable" either.

The D4 is the first DSLR that I'm aware of where you can
hold down the shutter release button and rattle off 100
shots in RAW mode without significant gaps while the
buffer is cleared.

I've never done that for a real shoot, I used to hit the
limit with a D3S on a regular basis, and never do with a
D4. Once you get used to a D4 everything else can be a
little annoying now and then, even a D800 with all those
fabulous pixels.

Just wait for the D5 generation... more pixels than a D800
and faster than a D4!

I will be deep into old fartdom by then (I suspect that will be true
for most of us here), and I believe I will probably have concerns other
than the burst rate provided by my DSLR. ;-)

--
Regards,

Savageduck

On 11/03/2013 10:31 AM, Floyd L. Davidson wrote:
Eric Stevens wrote:
On Mon, 11 Mar 2013 09:00:32 +1100, Rob wrote:

On 11/03/2013 7:56 AM, Eric Stevens wrote:
On Sat, 9 Mar 2013 19:45:01 -0800, Savageduck
wrote:

On 2013-03-09 19:20:01 -0800, Rob said:

On 10/03/2013 12:37 PM, Floyd L. Davidson wrote:
Wolfgang Weisselberg wrote:
At a back-of-the-envelope calculation, you'd need about 0.25
nm pixel size for f/1.2; which means about 96,000 x 144,000
pixels at FF, which comes to 13824 MPix (about 14 GPix).
Somehow 38 MPix seems a bit low compared to that.

In fact 150MP is getting close, and 250MP is certainly
enough that no anti-aliasing filter will be needed.
That is based on a diffusion limited lens with an f/1.2
aperture.

Hence something between 175 and 200MP is a pretty fair
estimate for the pixel count needed to simply eliminate
the need for any anti-aliasing filter on a 36x24mm
sensor.

Which is interesting, given that it would appear that
technology is just around the corner and will likely be
available in the next decade.



The next FF nikon looking likely to be 44Mp, now that's a worry,
thinking storage space.

Not to worry. It will come with its very own SATA SSD slot/dock.
...but no CF or SDHC slot. ;-)

How many seconds per shot on Continuous High?



That's not as funny as it sounds, if you have ever used the D100, were
they sloooow, the D200 became usable.

I didn't mean it to be funny. It's a serious problem.

And a lot of people didn't consider the D200 to be
"usable" either.


It was a big step up from the D100 buffer.




The D4 is the first DSLR that I'm aware of where you can
hold down the shutter release button and rattle off 100
shots in RAW mode without significant gaps while the
buffer is cleared.

I've never done that for a real shoot, I used to hit the
limit with a D3S on a regular basis, and never do with a
D4. Once you get used to a D4 everything else can be a
little annoying now and then, even a D800 with all those
fabulous pixels.

Just wait for the D5 generation... more pixels than a D800
and faster than a D4!

On Sun, 10 Mar 2013 16:12:53 -0700, Savageduck
wrote:

On 2013-03-10 15:37:01 -0700, (Floyd L. Davidson) said:

Savageduck wrote:
...and since the CF card is nothing but a slowish,
miniaturized SSD, why not develop an appropriately
dimensioned high performance SSD for pro-camera use?

It's called an XQD card.

Well, OK then!
So the future is sort of now, but so far only offered for the D4.
...and still a write speed of 145MB/s (1100x), which is basically the
same speed as the Lexar UDMA-7 at 145MB/s (1000x) at a cost difference
of $160. Why bother?

Note: $299 for a 32GB XQD, $579.99 for a 64GB XQD. Still about a
quarter the speed and well above the cost of equivalent capacity SSD
(60GB SSD @ $84 & 527MB/s write speed)
Still behind SSD's in both performance and cost/GB.

So my question remains, why not develop an appropriately dimensioned
high performance SSD for pro-camera use?

Because, at the moment, the card is not the bottle neck.
--

Regards,

Eric Stevens

On Sun, 10 Mar 2013 15:25:07 -0800, (Floyd L.
Davidson) wrote:

Eric Stevens wrote:
On Sun, 10 Mar 2013 14:48:42 -0700, Savageduck
wrote:

On 2013-03-10 13:56:05 -0700, Eric Stevens said:

On Sat, 9 Mar 2013 19:45:01 -0800, Savageduck
wrote:

On 2013-03-09 19:20:01 -0800, Rob said:

On 10/03/2013 12:37 PM, Floyd L. Davidson wrote:
Wolfgang Weisselberg wrote:
At a back-of-the-envelope calculation, you'd need about 0.25
nm pixel size for f/1.2; which means about 96,000 x 144,000
pixels at FF, which comes to 13824 MPix (about 14 GPix).
Somehow 38 MPix seems a bit low compared to that.

In fact 150MP is getting close, and 250MP is certainly
enough that no anti-aliasing filter will be needed.
That is based on a diffusion limited lens with an f/1.2
aperture.

Hence something between 175 and 200MP is a pretty fair
estimate for the pixel count needed to simply eliminate
the need for any anti-aliasing filter on a 36x24mm
sensor.

Which is interesting, given that it would appear that
technology is just around the corner and will likely be
available in the next decade.



The next FF nikon looking likely to be 44Mp, now that's a worry,
thinking storage space.

Not to worry. It will come with its very own SATA SSD slot/dock.
...but no CF or SDHC slot. ;-)

How many seconds per shot on Continuous High?


The 50GB buffer will handle that issue. ;-)

Well beyond the hypothetical fantasy consider the following:

If you consider that many SSD's currently have read/write specs at
559/527 MB/s, a 44MB file should be written in 0.0835 seconds. A 10
shot continuous burst should be captured in less than 1 second. It is
the buffer together with the card write speed which limits your
continuous high burst shot capacity.

You have to fill the buffer first and it is noticable that cameras
with large sensors are slower to fill the buffer than cameras with a
smaller sensor. I don't know where the bottle neck actually is but I
suspect that is the processing off the sensor.

Processing speed determines the maximum frame rate. The size of the
buffer compared to the size of each image determines how many shots
can be in a burst. The write speed to the memory card determines
how soon a consecutive burst can be started.

All of which is true. The question I was raising is about how rapidly
one can feed the buffer.

Given that you can currently buy a 480GB OWC SSD for $569, compared to
the fastest UDMA 32GB CF cards with write speeds of 145MB/s. at B&H
sale prices of $132.95. So even if compared to a 120GB SSD @ $150 the
equivalent capacity (let's say 4 x 32GB UDMA-7 CF cards, not an
unreasonable purchase for a pro) at a quarter the speed would cost that
pro $531.80.

...and since the CF card is nothing but a slowish, miniaturized SSD,
why not develop an appropriately dimensioned high performance SSD for
pro-camera use?

No doubt something like this will eventually happen but even now we
are pushing the limits in a number of respects.

The current XQD specification is for 500MB/s write
speeds, and XQD 2.0 specifications are 1000MB/s.

Do you have any data on the Continuous-High frame rate before the
buffer fills up on a real-world camera?

Wait ten years... when frame rates will be 24 fps,
images will be 200MP, and 500 shot bursts will be
possible. (Essentially high quality video for about 20
seconds.)
--

Regards,

Eric Stevens

On 10/03/2013 20:11, Robert Coe wrote:
[]
Fair enough, but one is still left with the central question: Is the alleged
benefit of a non-AA camera (a barely perceptible increase in sharpness of the
average picture) sufficient to motivate a prudent photographer to take the
risk? The orange blotches on the aforementioned jacket would seem to suggest
that the answer is "no". And the fact that one can envision a camera/lens
system that lowers that risk by some hard-to-determine amount, doesn't do much
to change that answer.

Bob

The orange blotches were not obtained with a 36 MP DSLR, I believe, so
are not representative of the 56 MP (full-frame equivalent) D7100. Have
we seen D7100 images where the problem actually shows? But I agree that
it's a judgement every prospective purchaser needs to make.
--
Cheers,
David
Web: http://www.satsignal.eu

On 10/03/2013 21:52, Floyd L. Davidson wrote:
David Taylor wrote:
On 10/03/2013 14:00, Floyd L. Davidson wrote:
[]
I implied no such thing. There are just different
effects with a 36MP sensor from having or not having an
anti-aliasing filter. Personally I chose to have the
filter, others choose not to. Each has it's upside, and
a downside too.

As I said, it's in part a personal choice.

The point was that 36MP is not high enough resolution
that diffraction effectively acts as an anti-aliasing
filter unless the aperture is extremely small. Note
that typical lenses are not even diffraction limited at
f/5.6, so the idea that such a lens would function as an
anti-aliasing filter is nonsense.

I am not suggesting that. At the Nyquist frequency if
the amplitude of the image components is sufficiently
low, the aliasing effects will similarly be of a low
amplitude, and may therefore be more tolerable.

You are erroneously claiming that a 36MP sensor puts the
Nyquist frequency above the resolving power of typical
lenses. The fact is that it doesn't even come close.

You apparently don't understand what the significance of
an anti-aliasing filter is, and is not.

Actually, I understand quite well.

If that were true you would not continue with the above
erroeous assertions.

One needs to
consider the total image chain (source characteristics,
atmospheric MTF, lens, focussing etc). rather than a
single perfect-lens/sensor calculation, to estimate how
bad aliasing may be in practice. We are moving towards
a point where the sensor will not be the limiting factor
in aliasing, and for some purposes that point has
already been reached. For you, perhaps not, and I'm
quite happy to accept that.

Yes, for some purposes it has been reached: shooting
pictures of very smooth blank walls that have zero
texture.

Do you do that often?

If you so much as have a fly land on that wall, it will
have detail above the Nyquist Limit.

The discussion is not about a 36 MP sensor, but one which has a pixel
density equivalent to a 56 MB full-frame one. To reach the resolving
limit of the lens requires (most likely) using a tripod, being in
perfect focus, having a subject which doesn't move significantly during
the exposure time, and having an atmospheric MTF which is perfect. Most
practical lenses will not reach the diffraction-limited resolution when
fully open.

Most of my photos fail on one or more of those criteria, if yours do not
then you will need a better sensor, I agree.

What is happening is that the sensor pixel density in the D7100 is now
at the levels where for the majority of users there is no need for an
anti-alias filter, or that's what Nikon appears to believe. I don't see
significant aliasing issues in these images, do you?

http://imaging.nikon.com/lineup/dslr/d7100/sample.htm

I await more sample images with interest.
--
Cheers,
David
Web: http://www.satsignal.eu

David Taylor wrote:
On 10/03/2013 21:52, Floyd L. Davidson wrote:
David Taylor wrote:
On 10/03/2013 14:00, Floyd L. Davidson wrote:
[]
I implied no such thing. There are just different
effects with a 36MP sensor from having or not having an
anti-aliasing filter. Personally I chose to have the
filter, others choose not to. Each has it's upside, and
a downside too.

As I said, it's in part a personal choice.

The point was that 36MP is not high enough resolution
that diffraction effectively acts as an anti-aliasing
filter unless the aperture is extremely small. Note
that typical lenses are not even diffraction limited at
f/5.6, so the idea that such a lens would function as an
anti-aliasing filter is nonsense.

I am not suggesting that. At the Nyquist frequency if
the amplitude of the image components is sufficiently
low, the aliasing effects will similarly be of a low
amplitude, and may therefore be more tolerable.

You are erroneously claiming that a 36MP sensor puts the
Nyquist frequency above the resolving power of typical
lenses. The fact is that it doesn't even come close.

You apparently don't understand what the significance of
an anti-aliasing filter is, and is not.

Actually, I understand quite well.

If that were true you would not continue with the above
erroeous assertions.

One needs to
consider the total image chain (source characteristics,
atmospheric MTF, lens, focussing etc). rather than a
single perfect-lens/sensor calculation, to estimate how
bad aliasing may be in practice. We are moving towards
a point where the sensor will not be the limiting factor
in aliasing, and for some purposes that point has
already been reached. For you, perhaps not, and I'm
quite happy to accept that.

Yes, for some purposes it has been reached: shooting
pictures of very smooth blank walls that have zero
texture.

Do you do that often?

If you so much as have a fly land on that wall, it will
have detail above the Nyquist Limit.

The discussion is not about a 36 MP sensor, but one
which has a pixel density equivalent to a 56 MB
full-frame one.

36MP, 56MP... or 100MP doesn't change the fact that
there will still be high frequency spatial data that is
above the Nyquist Limit of the sensor.

When the sensor reaches something above 175-200MP
that will change.

To reach the resolving limit of the
lens requires (most likely) using a tripod, being in
perfect focus, having a subject which doesn't move
significantly during the exposure time, and having an
atmospheric MTF which is perfect. Most practical lenses
will not reach the diffraction-limited resolution when
fully open.

That is not true. An *average* lens available today can
accomplish that. The differences between average lenses
and the best lenses are just a matter of the range over
which it can be accomplished.

Most of my photos fail on one or more of those criteria,
if yours do not then you will need a better sensor, I
agree.

Whether most of our shots fail to reach the maximum
potential of each and every part of our system is not
significant. If a particular job calls for a better
sensor today, it simply cannot be done.

What is happening is that the sensor pixel density in
the D7100 is now at the levels where for the majority of
users there is no need for an anti-alias filter, or
that's what Nikon appears to believe. I don't see
significant aliasing issues in these images, do you?

I don't think that is a correct evaluation of what
Nikon is doing at all. They try to sell what people
want to buy.

That doesn't mean Nikon is not aware that a 150MP
sensor will provide a better image in terms of aliasing.

http://imaging.nikon.com/lineup/dslr/d7100/sample.htm

I await more sample images with interest.

Do you actually expect Nikon marketing to provide images
that demonstrate a fault???

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