Some info on the 24 MP Nikon - from a Sony engineer

Started Oct 29, 2008 | Discussions
Vince P
Vince P Senior Member • Posts: 2,025
Re: Why wouldn't Nikon want to ADD a 24 mp camera if they could get the sensor?

headofdestiny wrote:

I doubt it. The A900 is testing very well in everything except ISO

over 1600.

The conclusion on the test here said that noise was a problem over 400 ISO. That doesn't detract from the fact it's a great camera and ties into some great lenses with the Carl Zeis optics.

 Vince P's gear list:Vince P's gear list
Sony RX100 VII Nikon D5 Nikon D850 Panasonic GH5S Nikon Z7 II +75 more
tissunique Veteran Member • Posts: 3,322
Re: Some info on the 24 MP Nikon - from a Sony engineer

I use the D2Hs which has the lbcast sensor and it's very good... but there's Kodak and others out there...
tony

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: Two inconsistencies...

Iliah Borg wrote:

Of course, it could be that Sony and Canon are better at designing
sensors

Wrong interpretation.

OK, if the right interpretation is that it might not be used in a
flagship

You do not follow.

No?
--
Bob

Alexramos
Alexramos Senior Member • Posts: 1,601
Re: expensive strategy

er1kksen wrote:

Alexramos wrote:

NOTE: is very funy read how people say
that K20D sensor is a design of Pentax/Samsung... please, Pentax is
like Nikon, they don´t design alone sensor.

This is getting kind of annoying, and also this: "the 14.something MP
cropped sensor that's in a couple of Sonys and Pentaxes"

It's also very funny to read about how Samsung is the largest
electronics company in the world, larger than Sony, and that they do
in fact make their own sensors. The 6mp and 10mp sensors in previous
cameras have been Sony designs, but the 14.6mp sensor in the K20D is
in fact a Samsung design, having nothing to do with Sony, and
especially nothing to do with the rather mediocre 13.5mp sensor in
the Sony A350.

This information is readily available to anyone who looks. It sounds
like some of the posters here seem to think that Sony and Canon are
the only companies making sensors, one post even seeming to imply
that canon uses sony sensors (it was difficult to descipher, however,
so I may be wrong). This is simply not the case: As pointed out,
Kodak and Fujifilm make sensors, Panasonic makes all the current 4/3
sensors, Samsung has put out the K20D sensor as well as many sensors
for video, and Nikon themselves have stated that Sony did not make
the D3 sensors. Some individuals with a bit of insider knowledge have
implied that a smaller sensor-manufacturing company located in japan
which otherwise doesn't have any presence in the photography
industry. And yes, they do have the experience to design sensors. A
corporation is not a singular entity. It's a group, and they can hire
the people to do the job if they don't have the know-how in the first
place. They've been designing the sensors for their flagship cameras
from the D2H up.

Clearly, some of the posters here know what they're talking about
quite well, and I fear they may be the only ones who read this, and
they're not the ones who need to.

Learn before you post!

Sure, the GX-20/H20D sensor is a Samsung sensor and don´t a Pentax sensor, but you are wrong...

Samsung is not longer than Sony, but it is not important Samsung is a great electronic company with a very good products.

The 14MPx on a350 is a CCD, very different to 14MPx CMOS sensor on GX-20 in noise, but the noise is not the onlyone feature of a sensor.
Has you compare both sensor to ISO 100-400 in resolution, details and sharpen??

The a350 sensor is as so sharp like a100 sensor, the most sharpen and details sensor CCD until now, a new version of D200 CCD sensor and the same on D80 camera, but with a different design in microlenses.

The Samsung sensor has the same size of Sony sensor and both company make about 50% of all APS-C sensor on digital camera now (Samsung, Pentax, Sony, Nikon).

The D2H sensor was made for Kodak and don´t designed for nikon along.
Renesa is very posible the maker of D3 sensor.

Sure, Kodak, Panasonic, Fuji (they don´t make their sensors), and Foveon are other company maker of DSLR sensor. Kodak now is not make sensor for DSLR just for S2 is the last.

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: expensive strategy

Alexramos wrote:

Is many probably that CMOS 12MPx APS-C was a agreemend between Sony
and Nikon for a sustitute of D2x and this sensor had some time in
design. Maybe then Nikon choose FF for D3 and used the CMOS 12MPx on
D300.

There are probably very many agreements, since Nikon is Sony Semi's major customer in the DSLR sensor market. Who know's what's happening. I wouldn't be surprised to see a D60x with the A350 sensor, for instance.

Sony from 10MPx CCD haven´t sold more DSLR sensor and is logic, Sony
is in DSLR market.

The logic is more that their customers have gone elsewhere. They OEMed to Nikon, KM and Pentax. Pentax is now supplied by Samsung, the other two remain, although KM is part of Sony. I would uess that the Nikon roadmap doesn't include any more CCD, and the IMX021 woill find its way down to the low end next year

The D300/D90 sensor is the same on a700 and, why the first has 14bits
and the second just 12bits??
The 14bits on D300 sensor I think is just a exposure time
manipulation, but the native tone gamma is 12bits.

There are several theories. I think the D300 simply reads the data four times and adds the result to give a 14 bit sample.

Nikon has a big problem:
-Don´t has other APS-C sensor now

I suspect that the consumer road map is based on the IMX021. Nikon will shift of that sensor than Sony.

-Don´t has other FF sensor now

You know what's in the Nikon development labs?

-Nikon don´t has the experience for design a sensor alone or built it
(How can Nikon has experience and make a better sensor if never,
never, they have made one)

Nikon has made two, the D2h sensor and the D3/D700 sensor. If you don't believe that it made by Nikon, then why does it say 'Nikon' on the chip?

NOTE: is very funy read how people say
that K20D sensor is a design of Pentax/Samsung... please, Pentax is
like Nikon, they don´t design alone sensor.

In that respect Pentax is not like Nikon, since Nikon has designed sensors alone.

-They are looking for a new supplie (Microsoft??, MF maker??)

Nikon has a problem similar to the new sensor, BIG. They used the
same D300 sensor on D90 because doesn´t has other.

Or else because 12.3 MPix was a suitable spec for the camera, and fitting it in the higher volume camera will result in economies of scale.

They will use the same D300/D90 sensor on the D60 sustitute and the
D700 was a escape way for its market.

Very likely that sensor will find its way to the low end. What has the D700 got to do with it?

The FF 12MPx CMOS sensor on D3/D700 is good?? sure, is a excelente
sensor, but it can´t be compared with FF 24MPx CMOS sensor on a900.

Some aspects can be compared. Obviously, it is a different sensor for a different purpose.

The quality and performance of Nikon FF sensor is for pixel size and
ISO dedicate and the Sony sensor is for resolution, details and
sharpen image.

The Nikon sensor is 1/3 stop more efficient, and that has nothing to do with pixel pitch.

The FF Sony sensor has a very good image quality to high ISO, is very
similar until ISO 3200 to D3/D700 sensor from RAW image.

It is much better than it gets credit for. However, low ISO performance is not sensor limited, rather it's read chain noise limited. It's at high ISO where the intrinsic efficiency of the sensor shows.

Nikon know and used for many years Sony sensor and to say now that
Nikon doesn´t want the Sony sensor because it is 12bits or has not
the quality, is stupid, Nikon need the Sony sensor and ever they have
used them, but now the market and history is defferent and them
sensor are not available.

That's one point of vie, just not a very sensible one.

Look the Sony forum or web site of Minolta/Sony and you will see to
many Nikon user asking for a900 and buying it. The a900 is a good
point in the actual market and a good solution for Nikon user that
want a high resolution DSLR and D3x never came.

Some people have the money and don't ant to wait. Some are looking seriously at the 5DII, as well. Nikon does need a product in this place quite quickly.

Saving for Sony 70-300SSM G
I would like change my a100 for a350...

Brand loyalty wins of rational discussion, again.
--
Bob

Vince P
Vince P Senior Member • Posts: 2,025
Re: expensive strategy

bobn2 wrote:

The D300/D90 sensor is the same on a700 and, why the first has 14bits
and the second just 12bits??
The 14bits on D300 sensor I think is just a exposure time
manipulation, but the native tone gamma is 12bits.

There are several theories. I think the D300 simply reads the data
four times and adds the result to give a 14 bit sample.

No sensor has any bits as we understand it. Well technically on a quantum level it does but effectively it's analog. It's the A/D convertor that determines how many bits you get in your memory card. The A/D convertors on the D300 and A700 are completely different so that's why they are different.

 Vince P's gear list:Vince P's gear list
Sony RX100 VII Nikon D5 Nikon D850 Panasonic GH5S Nikon Z7 II +75 more
er1kksen New Member • Posts: 7
Re: expensive strategy

Yes, Alex, Samsung is indeed larger than Sony. Go look it up. Not that it really matters, the point was that Samsung certainly has the capability to produce sensors, and like Nikon, Pentax may well have some of their own ideas about sensor design which may be incorporated.

I have certainly compared the two sensors, and of course noise isn't the only difference. Other differences are that the images from the Samsung sensor contain more detail (not a lot more, but some), better color, and smoother tonal rendering, at least to my eye. This is at low ISO, of course. And of course it's the same size as the Sony sensor, it's APS-C.

Alexramos wrote:

er1kksen wrote:

Alexramos wrote:

NOTE: is very funy read how people say
that K20D sensor is a design of Pentax/Samsung... please, Pentax is
like Nikon, they don´t design alone sensor.

This is getting kind of annoying, and also this: "the 14.something MP
cropped sensor that's in a couple of Sonys and Pentaxes"

It's also very funny to read about how Samsung is the largest
electronics company in the world, larger than Sony, and that they do
in fact make their own sensors. The 6mp and 10mp sensors in previous
cameras have been Sony designs, but the 14.6mp sensor in the K20D is
in fact a Samsung design, having nothing to do with Sony, and
especially nothing to do with the rather mediocre 13.5mp sensor in
the Sony A350.

This information is readily available to anyone who looks. It sounds
like some of the posters here seem to think that Sony and Canon are
the only companies making sensors, one post even seeming to imply
that canon uses sony sensors (it was difficult to descipher, however,
so I may be wrong). This is simply not the case: As pointed out,
Kodak and Fujifilm make sensors, Panasonic makes all the current 4/3
sensors, Samsung has put out the K20D sensor as well as many sensors
for video, and Nikon themselves have stated that Sony did not make
the D3 sensors. Some individuals with a bit of insider knowledge have
implied that a smaller sensor-manufacturing company located in japan
which otherwise doesn't have any presence in the photography
industry. And yes, they do have the experience to design sensors. A
corporation is not a singular entity. It's a group, and they can hire
the people to do the job if they don't have the know-how in the first
place. They've been designing the sensors for their flagship cameras
from the D2H up.

Clearly, some of the posters here know what they're talking about
quite well, and I fear they may be the only ones who read this, and
they're not the ones who need to.

Learn before you post!

Sure, the GX-20/H20D sensor is a Samsung sensor and don´t a Pentax
sensor, but you are wrong...
Samsung is not longer than Sony, but it is not important Samsung is a
great electronic company with a very good products.
The 14MPx on a350 is a CCD, very different to 14MPx CMOS sensor on
GX-20 in noise, but the noise is not the onlyone feature of a sensor.
Has you compare both sensor to ISO 100-400 in resolution, details and
sharpen??
The a350 sensor is as so sharp like a100 sensor, the most sharpen and
details sensor CCD until now, a new version of D200 CCD sensor and
the same on D80 camera, but with a different design in microlenses.
The Samsung sensor has the same size of Sony sensor and both company
make about 50% of all APS-C sensor on digital camera now (Samsung,
Pentax, Sony, Nikon).

The D2H sensor was made for Kodak and don´t designed for nikon along.
Renesa is very posible the maker of D3 sensor.

Sure, Kodak, Panasonic, Fuji (they don´t make their sensors), and
Foveon are other company maker of DSLR sensor. Kodak now is not make
sensor for DSLR just for S2 is the last.

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: expensive strategy

Alexramos wrote:

The D2H sensor was made for Kodak and don´t designed for nikon along.
Renesa is very posible the maker of D3 sensor.

Could you cite the source of your information that the D2h sensor was made 'for Kodak'. If so, it would appear that Nikon has a court case coming its way for all the false iformation they put out about the design of the LBCAST sensor. Nikon is the maker of the D3 sensor. There is no clear information of which foundry the silicon was processed in, although some have suggested Renesas - so far as I can see, this is based on no complelling evidence.

Sure, Kodak, Panasonic, Fuji (they don´t make their sensors)

That's interesting information, but wrong. Fujifilm do make their sensors, but since last year they have subcontracted wafer processing to Toshiba. I think Toshiba is at least as likely as Renesas to be undertaking wafer processing for Nikon also. since they already do it for Sony, that would be an interesting. Kodak's CMOS sensors are fabricated by IBM, in a fab which is process compatible with Toshiba and Sony. Complex, isn't it.

, and

Foveon are other company maker of DSLR sensor.

Foveon is also 'fabless', their wafer processing is done by Dongbu (I bet you've never heard of them! - one of the largest CIS oundries in the world, maybe the do the Nikon wafers!)
--
Bob

FrankG Senior Member • Posts: 2,242
Re: expensive strategy

bobn2 wrote:

Tiffles wrote:

That is going to be one expensive sensor for Sony then - wonder how
much they think they will sell in their own bodys. Also, don't Sony
and Nikon have a long business relation in terms of 'sensor pooling'?

How about this pure imaginary bit: Nikon wasn't happy with the sensor
Sony made (no movie function, live view, noisy, in sum worse than
what the new Canon sensor is hoped to be) and now have to design one
themselves.

Well I think Nikon may be considering other options, but not for
those reasons. The A900 sensor is pretty much an upscale of the
sensor in the D300 and D90. With that they manage movie function and
live view. The noise is also not nearly as bad as made out. At equal
output sizes, it seems to be in the same ballpark as the D700, by
comparisons done on this forum, just not quite there (in fact, pretty
much the 1/3 stop worse I suggested). If people insist on looking at
high density cameras at 100% crop, then they're always going to go on
about the noise.
--
Bob

Even at base ISO?

If you were right then what would be the point of having such a "higher density" (as you put it) camera if you end up having to downsize the output to make it acceptable noisewise? If such is the case then there isn't any point at all.

Of course I would actually say that there is a point for at least some applications such as slow landscape and maybe portrait and studio work at low ISOs where you should easily get clean output from 24Mpix FF cameras.

However as for the A900 being only "one third of a stop" worse noise than the D700 "at equal output sizes" - would you mind pointing to your evidence for that?

  • Frank

 FrankG's gear list:FrankG's gear list
Nikon D500 Nikon D850 Nikon Z50
frozenlight Contributing Member • Posts: 583
Re: How many PROs use Sony? Any Pulitzers shot with a Sony?

NikonProArtist wrote:

frozenlight wrote:

Gosh, Nikon must be terrified!

Amen!

This discussion is interesting from a theoretical perspective, but
that is all. In the end it won't make one iota of difference whether
or not Sony sells their sensor to anyone else. If anything, Sony is
in such sad financial shape these days they have more reason to sell
products than ever before.

Exactly.

As for the the Pulitzer, from what I recall reading about a year or
two ago 65 of the 83 Pulitzer prize winning photos were shot by
photographers using Nikon equipment. That being said, who in their
right mind believes that photographers en masse are going to give up
their Nikon gear and replace it with Sony gear?

My guess... precisely ZERO.

-- hide signature --

Sensorly yours...

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: expensive strategy

Vince P wrote:

bobn2 wrote:

The D300/D90 sensor is the same on a700 and, why the first has 14bits
and the second just 12bits??
The 14bits on D300 sensor I think is just a exposure time
manipulation, but the native tone gamma is 12bits.

There are several theories. I think the D300 simply reads the data
four times and adds the result to give a 14 bit sample.

No sensor has any bits as we understand it. Well technically on a
quantum level it does but effectively it's analog. It's the A/D
convertor that determines how many bits you get in your memory card.
The A/D convertors on the D300 and A700 are completely different so
that's why they are different.

Since the IMX021 sensor, which is very likely the one in the D300, has on board analog to digital coversion, and 12 bit outputs, it does in a real sense have 'bits' as we understand it. There is no hard information in the public domain that the A/D converters on the D300 and A700 are completely different, since Nikon has not released any. The Sony IMX021 documentation makes no mention of any 14 bit capability, and the 'four read' scenario fits the observed characteristics of the D300 14 bit mode, in my opinion. Others have other opinions, but no-one has hard information (unless you are claiming differently).
--
Bob

frozenlight Contributing Member • Posts: 583
Re: Surprising numbers, and you'd be surprised...

Joseph S Wisniewski wrote:

frozenlight wrote:

Re: How many PROs use Sony?

Well, if you consider the number who have been using Minolta AF for
decades, starting back when Nikon's AF strategy was, well, a little
"confused"...

And continuing to when Minolta brought out the 9000, a monster with
the fastest drive speed, top shutter speed, and top sync speed on the
pro market...

And those who shot portrait with the 135mm SF, arguably the most
interesting "controlled bokeh" and "soft focus" portrait tele (using
an apodizing disc instead of the Nikon and Canon approach of
manipulating SA)...

A quite surprising number.

OK so surprise me... name some published pros using Sony DSLR's...

Any Pulitzers shot with a Sony?

They've got both the Minolta and Konica heritage. Four Pulitzers just
with the X370, that's just one Minolta model, even before the Maxxum,
and a few for the Konica Hexar. Now, the Autoreflex-T, that's
something that probably should remain buried. With a cross nailed to
the inside of the coffin lid, to keep it from rising again...

OK. ZERO Sony DSLR's as I know. 60+ with Nikon. QED.

Gosh, Nikon must be terrified!

Of course not. But that's not the issue, is it?

Right. This whole issue is bogus. Sony will sell to anyone because they need economies of scale. Nikon may prefer to design their own chip and get Sony to fab it (as they have done in the past).

-- hide signature --

Rahon Klavanian 1912-2008.

Armenian genocide survivor, amazing cook, scrabble master, and loving
grandmother. You will be missed.

Ciao! Joseph

http://www.swissarmyfork.com

-- hide signature --

Sensorly yours...

Haplothedog Regular Member • Posts: 119
Re: Two inconsistencies...

bobn2 wrote:

Iliah Borg wrote:

A good reason for Nikon not to use the A900 sensor in a flagship
model is that it has become better at designing sensors than Sony (or
Canon, for that matter).

Interesting how you manage to have two assumptions in one sentence

Yes, but there is some evidence to back the assumptions up. If you
have counter evidence...
Of course, it could be that Sony and Canon are better at designing
sensors but have decided for some reason not to release their better
sensors.
--
Bob

Bob, you keep on saying that on DPR, but if we look at Roger N. Clark. (www.clarkvision.com) comprehensive work on Digital Camera Sensor Performance, it appears that (if I am not mistaken, any reader should visit clarkvision for details), comparing Nikon D3 and Canon 1DIII :

  • 1D III full well capacity is better than D3, even with smaller pixel pitch

  • 1D III Signal-to-Noise Ratio is better than D3, even with smaller pixel pitch

  • 1D III Read Noise is better than D3

  • 1D III Dynamic Range is better than D3

  • D3 Low Light Sensitivity Factor (high iso shadow perf) is better than 1DIII

  • D3 apparent IQ is better than 1DIII, mainly due to larger pixels.

Mr Clark also proposes another view of sensors, using megapixels * pixel pitch and so we have :
1. Sony A900
2. Canon 1Ds Mark III & Canon 5D Mark II
4. Nikon D3
5. Canon 1D Mark III

Maybe Nikon better designing is not that evident ?

Regards, Francois

avidday Senior Member • Posts: 2,082
Re: expensive strategy

Vince P wrote:

The A/D convertors on the D300 and A700 are completely different so
that's why they are different.

They are not "completely different". The A700 and D300 are using the same basic "IMX021" sensor which uses column parallel delta encoded A/D converters arrays fabbed directly into the sensor silicon. At best, the only difference with the D700 is the ability to let the A/D converters quantize in 14 bits. For delta encoded ADCs that might require no additional logic at all, because they are basically just counters, and to quantize to a higher bit count only requires more counting (given there is enough accumulator and counter register width and DAC precision). It is also completely consistent with the reduced frame rate the D300 can achieve in 14 bit mode.

If there are differences between the D300 and D700 quantization circuits, all indications are that they range between minimal and non-existent.

Some more about the sensor can be found here.

http://www.sony.net/SonyInfo/News/Press/200708/07-072E/index.html

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: expensive strategy

FrankG wrote:

bobn2 wrote:

Tiffles wrote:

That is going to be one expensive sensor for Sony then - wonder how
much they think they will sell in their own bodys. Also, don't Sony
and Nikon have a long business relation in terms of 'sensor pooling'?

How about this pure imaginary bit: Nikon wasn't happy with the sensor
Sony made (no movie function, live view, noisy, in sum worse than
what the new Canon sensor is hoped to be) and now have to design one
themselves.

Well I think Nikon may be considering other options, but not for
those reasons. The A900 sensor is pretty much an upscale of the
sensor in the D300 and D90. With that they manage movie function and
live view. The noise is also not nearly as bad as made out. At equal
output sizes, it seems to be in the same ballpark as the D700, by
comparisons done on this forum, just not quite there (in fact, pretty
much the 1/3 stop worse I suggested). If people insist on looking at
high density cameras at 100% crop, then they're always going to go on
about the noise.
--
Bob

Even at base ISO?

It's tricky to judge at base ISO, since the available sensor DR of decent camera so far exceeds the DR of most available output media, that they all end up looking much the same, even if the measure differently. I haven't done any measurements, but I get the impression (from such things as the DPR DR test, for instance) that the A900 has quite good low ISO read noise, and therefore quite extended low iSO DR.

If you were right then what would be the point of having such a
"higher density" (as you put it) camera if you end up having to
downsize the output to make it acceptable noisewise? If such is the
case then there isn't any point at all.

No-one said anything about downsizing the output. We invariably upsize it, since no-one's keen looking at images 24x36 mm. The issue is, what do you lose by increasing pixel density? The answer is nothing (except file size and processor capacity). If you choose to make an image to your prefences with a low and high pixel density camera with the same size and tech sensor, the noise will be of the same amplitude, so you haven't lest anything. What you've gained is the ability to make larger prints with some detail should you choose. Obviously, when doing so you may choose not to go for extreme ISO's or accept the noise or go B&W, but its an option you don't have with the smaller image size. Moreover, the higher pixel density image will have more detail even at smaller image sizes, even if the advantage is small at small sizes.

Of course I would actually say that there is a point for at least
some applications such as slow landscape and maybe portrait and
studio work at low ISOs where you should easily get clean output from
24Mpix FF cameras.

The great thing about high pixel density is you can choose. low noise? Don't print so big. Big print? Stick to 'low' iso. How often do you really use ISO's greater than 400 anywa? It won't be 'never' but it certainly won't be all the time, either, unless you're doing very specialist stuff.

However as for the A900 being only "one third of a stop" worse noise
than the D700 "at equal output sizes" - would you mind pointing to
your evidence for that?

First look at Emil Martinec's tables ( http://theory.uchicago.edu/~ejm/pix/20d/posts/tests/D300_40D_tests/ ) - the D300 has to stand in for the A900, and adjust for area. This might not be completely accurate, but since sensor efficiency seems to go across sensor families (look at the Canon ones) it's a fair assumption. Now look at Clarkvision ( http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html ) again letting the D300 stand in for the A900. On the sensor efficiency side (from Emil), we find that the D3(or D700) has about 1/3 stop advantage in area efficiency. Since the sensors are the same size, that translates to overall efficiency. Now look at the read noise figures (from Roger) and we find an advantage (more than a stop) to the A900 at low ISO's, with them being about even at high ISO's. This suggests that if anything the A900 will have a slight advantage at low ISO's. this actually triangulates with real test results, where the A900 seems to have a measurable DR advantage over the D3/D700.

It's a bit more difficult to find noise tests in raw at high iSOs (and JPEG tests tell you nothing about the sensor qualities) but this thread here: http://forums.dpreview.com/forums/readflat.asp?forum=1037&thread=29804929 did an equal size comparison in RAW at 3200 and I'd agree with his conclusion, the A900 holds up quite well, certainly seems to triangulate with my (around 1/3 stop) estimate. Above 3200 we'd expect the D3/D700 to pull further ahead, since its sensor efficiency advantage is still there and relative read noise is dropping. Beware of basing you assessment on JPEGs, the A900 seems to have a very poor JFIF engine. That may affect your purchase decisions, but doesn't tell you much abour relative sensor efficiency.

-- hide signature --

Bob

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: Two inconsistencies...

Haplothedog wrote:

bobn2 wrote:

Iliah Borg wrote:

A good reason for Nikon not to use the A900 sensor in a flagship
model is that it has become better at designing sensors than Sony (or
Canon, for that matter).

Interesting how you manage to have two assumptions in one sentence

Yes, but there is some evidence to back the assumptions up. If you
have counter evidence...
Of course, it could be that Sony and Canon are better at designing
sensors but have decided for some reason not to release their better
sensors.
--
Bob

Bob, you keep on saying that on DPR, but if we look at Roger N.
Clark. (www.clarkvision.com) comprehensive work on Digital Camera
Sensor Performance, it appears that (if I am not mistaken, any reader
should visit clarkvision for details), comparing Nikon D3 and Canon
1DIII :

  • 1D III full well capacity is better than D3, even with smaller

pixel pitch

Yes, but you are comparing the 1DIII FWC at 100ISO with the D3 at 200ISO. The D3 gathers more photoelectrons per unit area than the 1DIII at 200ISO, and retains that advantage all the way up. It also has a big ISO 200 FWC advantage (65,600 plays 35,200, assuming that ISO 200 is simply ISO 100 with twice the read chain gain) at 200ISO. It's not as good at 100ISO. This is a design decision. restrict the cell capacitance, you get a lower FWC which translates to a higher conversion gain.

  • 1D III Signal-to-Noise Ratio is better than D3, even with smaller

pixel pitch

Only at 100ISO. If we define SNR as FWC/read noise (which is one sensible measure) then at 200, the 1DIII gets 35200/12.2 = 2869 = 11.5 stops whereas the d3 gets 65600/17.6 = 3727 = 11.9 stops. At 1600 we get 4400/4.2 = 1048 = 10 stops for the 1DIII and 8200/4.9 =1673 = 10.7 stops for the D3. Then, of course, the D3 has 50% more sensor area, which compounds the advantage. This triangulates with the real test results showing the D3 to have a clear advantage at high ISO's. The 1DIII has a very slight advantage in the 'best possible' from the camera. at 100ISO, it gets 70200/24.4 = 2877 = 11.5 stops against the D3's best at 200ISO. However, since the D3 has the advantage that its sensor is 1.5 times larger, it beats the 1DIII comfortably. It's not so much better off against the 1DsIII, where it loses the advantage. Overall, given the D3's better sensor efficiency and the 1DsIII's better read chain, the result seems to be a toss up, depending on sample variation as much as anything.

  • 1D III Read Noise is better than D3

Normalised to base ISO (100 plays 200) the D3 is actually better.

  • 1D III Dynamic Range is better than D3

As shown above, only at 100 ISO and then very marginally.

  • D3 Low Light Sensitivity Factor (high iso shadow perf) is better

than 1DIII

As shown above

  • D3 apparent IQ is better than 1DIII, mainly due to larger pixels.

I'm not sure of the value of this measure, but in any case, it's due to more pixels AND a larger sensor.

Mr Clark also proposes another view of sensors, using megapixels *
pixel pitch and so we have :
1. Sony A900
2. Canon 1Ds Mark III & Canon 5D Mark II
4. Nikon D3
5. Canon 1D Mark III

That takes no account of the actual quality of the sensor as a device for converting light to electric signals.

Maybe Nikon better designing is not that evident ?

Completely evident on those figures, if you read them better. As i said, Nikon has the lead in sensor design. Canon is still clearly better at read chain design. Put the two together, and it would be quite a camera. Since i mentioned it, using the d3 sensor with the 1DIII read chain you'd get

65600/12.2 = 5377 = 12.4 stops at 200ISO and 8200/4.2 = 1952 = 10.9 stops at 1600 ISO. That's half a stop advantage over the D3 at 200, alas only 0.2 stops at 1600.

-- hide signature --

Bob

bobn2
bobn2 Forum Pro • Posts: 69,742
Re: expensive strategy

avidday wrote:

Vince P wrote:

The A/D convertors on the D300 and A700 are completely different so
that's why they are different.

They are not "completely different". The A700 and D300 are using the
same basic "IMX021" sensor which uses column parallel delta encoded
A/D converters arrays fabbed directly into the sensor silicon.

They aren't delta encoders, they're single slope converters.

At
best, the only difference with the D700 is the ability to let the
A/D converters quantize in 14 bits. For delta encoded ADCs that might
require no additional logic at all, because they are basically just
counters, and to quantize to a higher bit count only requires more
counting (given there is enough accumulator and counter register
width and DAC precision).

But since they aren't delta converters it would require different silicon, a 14 bit DAC for the slope generator and 14 bit count latch registers in each column.

It is also completely consistent with the
reduced frame rate the D300 can achieve in 14 bit mode.

So is sampling four times.

If there are differences between the D300 and D700 quantization
circuits, all indications are that they range between minimal and
non-existent.

I vote for non-existent. They are too large to be a simple mask option.

Some more about the sensor can be found here.

http://www.sony.net/SonyInfo/News/Press/200708/07-072E/index.html

-- hide signature --

Bob

Roger Senior Member • Posts: 2,830
From what I understand Fuji

Greetings

For what it's worth, I understand that Nikon will be making a announcement later this year stating they will be buying Fuji's DSLR and chip division or buying into it. If this is true, problem solved, good bye Sony.

Enjoy
Roger J.

glacierpete Senior Member • Posts: 1,917
Re: He didn't say they don't want a 24mp camera...

The Sony sensor does not look bad at all. A900 +CZ24-70 compared to Hasselblad H1-P30
http://forums.dpreview.com/forums/read.asp?forum=1037&message=29852201

Iliah Borg Forum Pro • Posts: 28,632
Re: Two inconsistencies...

Yes, but you are comparing the 1DIII FWC at 100ISO with the D3 at
200ISO.

On the sensor level there is no such thing as ISO. On the camera level ISO is a pretty arbitrary number.

-- hide signature --
Keyboard shortcuts:
FForum MMy threads