Imaging Resource takes a look inside Nikon's sensor design studios, solves Sony misconceptions

[Andrew Reid]

Dave Etchells has a background in chip science and finally had the very rare opportunity to visit Nikon's sensor design labs. It's long been the perception in the photographic and filmmaking community that Nikon leans heavily on Sony for off-the-shelf sensors and is handed down parts. It's led to opinions like "why would Sony give their best sensor to rival Nikon?" being quite common on camera sites and forums.

The answer, in actual fact, is that Nikon can beat Sony at their own game, with the 46MP full frame sensor in the D850 proof of the pudding. It does better 4K video quality than the A7R III at a higher megapixel count and edges it for stills quality too. Now let's hear about some of the science behind it.

Read the full article

[Nikkor]

Nice post.

[Robert Collins]

You can be sure that if Nikon makes their own sensors for the upcoming FX Z-mount mirrorless cameras, they will be true Nikon custom silicon and not just a “greatest hits” of other people’s technology. Indeed it might not even matter if they are manufactured in facilities owned by Sony, Toshiba or TowerJazz – it is the design process which is key to defining the performance, both for stills and video. Quote...

Honestly, this is missing the whole point.

Pretty much anyone with say well less than US$100m investment can 'design' an incredibly fancy sensor with 'adc', 'BSI', 'stacked sensor technology' '10nm'. Now go out there and find yourself a 'foundry' to manufacture it for you and you will find there is exactly 'one' manufacturer who can do it - and that is 'Sony'. (Well possibly Samsung can do it but doesnt seem interested.)

I read about an estimate for say Canon to upgrade their silicon manufacture to Sony's and it was something like US$16bn.....

(PS. Incidentally Sony bought out Toshiba a year ago.)

I am unsure whether it is Sony's tech (such as stacked cmos sensors) or the incredible 'capital investment' (which in Sony's case is supported by its smartphone sensors) which is the key to defining your sensor performance - but it certainly isnt the 'design'. Designers are 10 a penny.

[Andrew Reid]

How do you explain why the Nikon design out-performs the Sony from the same foundry then?

[MdB]

19 minutes ago, Robert Collins said:

You can be sure that if Nikon makes their own sensors for the upcoming FX Z-mount mirrorless cameras, they will be true Nikon custom silicon and not just a “greatest hits” of other people’s technology. Indeed it might not even matter if they are manufactured in facilities owned by Sony, Toshiba or TowerJazz – it is the design process which is key to defining the performance, both for stills and video. Quote...

Honestly, this is missing the whole point.

Pretty much anyone with say well less than US$100m investment can 'design' an incredibly fancy sensor with 'adc', 'BSI', 'stacked sensor technology' '10nm'. Now go out there and find yourself a 'foundry' to manufacture it for you and you will find there is exactly 'one' manufacturer who can do it - and that is 'Sony'. (Well possibly Samsung can do it but doesnt seem interested.)

I read about an estimate for say Canon to upgrade their silicon manufacture to Sony's and it was something like US$16bn.....

(PS. Incidentally Sony bought out Toshiba a year ago.)

I am unsure whether it is Sony's tech (such as stacked cmos sensors) or the incredible 'capital investment' (which in Sony's case is supported by its smartphone sensors) which is the key to defining your sensor performance - but it certainly isnt the 'design'. Designers are 10 a penny.

This. 100% this. 

This whole article about the design process at Nikon is literally designed to make people think they have more in house involvement than they do. 

Read between the lines and Nikon do the following:

1. Design the way the chips are packaged

2. Design sensor toppings 

Thats about it. Nikon are not designing a sensor from scratch and saying 'hey make this'. Sony give them the basics of what can be made and Nikon customise. 

Isnt it funny they say they 'design down to xyz' and then point at the bits we already know they (and everyone else making big enough orders) customise? 

There is no secret sauce here. D850 sensor is better than A7R III sensor? Really? Sure about that? A7R III achieves the near enough same DR but at ISO 100 rather than 64. A7R sensor has OSPDAF that the Nikon lacks. They are similar enough and yet the Nikon sensor is what, 2-3 years newer and still doesn't have the tech in the Sony. 

I would be very interested to see where Nikon is getting its mirrorless sensors from. 

11 minutes ago, Andrew Reid said:

How do you explain why the Nikon design out-performs the Sony from the same foundry then?

Which sensor specifically? Sony rarely uses the 'same' sensors as their off the shelf counterparts. Those that are the actual same units usually perform (according to DXO etc) as close enough to identical (or other factors that explain any minor differences).

Explain this: Why do Sony get to sell 'Nikon designed' sensors to Pentax? Or Fuji? Or anyone else that will buy them? 

[Andrew Reid]

Oh dear. Is it ignore the articles and talk out of your ass day again?

[Robert Collins]

8 minutes ago, Andrew Reid said:

How do you explain why the Nikon design out-performs the Sony from the same foundry then?

Honestly, I dont think it does....

When you 'design' a sensor your have to make trade offs. If you choose a 'low base iso' you will score 'higher' than a sensor in 'dynamic range' than a sensor with a 'higher base iso'. But you will score 'lower' in terms of 'high iso capability'.

So take the D850 v A7riii

https://www.dxomark.com/Cameras/Compare/Side-by-side/Sony-A7R-III-versus-Nikon-D850___1187_1177

The D850 scores better for DR (because its base iso is 64), the A7riii higher for iso performance. The fact that the Nikon sensor 'outperforms' Sony in dynamic range doesnt really mean it outperforms it as a sensor.

But even if you argument is that Nikon is 'better at designing sensors' than 'Sony', it doesnt get around the fact that 'Sony' is better at manufacturing sensors than anyone else!!

[MdB]

2 minutes ago, Andrew Reid said:

Oh dear. Is it ignore the articles and talk out of your ass day again?

Oh so Nikon didn't invite these journos in to try and spin what they do? Do you really think that's the case? There has been spin from the Nikon world ever since they stopped making their own sensors to try and state that they are their own work. Why do you think they always obscure who actually makes them? 

2 minutes ago, Robert Collins said:

Honestly, I dont think it does....

When you 'design' a sensor your have to make trade offs. If you choose a 'low base iso' you will score 'higher' than a sensor in 'dynamic range' than a sensor with a 'higher base iso'. But you will score 'lower' in terms of 'high iso capability'.

So take the D850 v A7riii

https://www.dxomark.com/Cameras/Compare/Side-by-side/Sony-A7R-III-versus-Nikon-D850___1187_1177

The D850 scores better for DR (because its base iso is 64), the A7riii higher for iso performance. The fact that the Nikon sensor 'outperforms' Sony in dynamic range doesnt really mean it outperforms it as a sensor.

But even if you argument is that Nikon is 'better at designing sensors' than 'Sony', it doesnt get around the fact that 'Sony' is better at manufacturing sensors than anyone else!!

And this. There was so much chest thumping associated with the D850 about how perfect it was because it had to be (rather than it is) that a lot of the 'facts' were decided about the camera before the thing even landed in anyone's hands. 

[Andrew Reid]

The D850 46MP sensor does outperform the latest Sony 42MP. It is a newer design, Sony has tweaked an old design for the latest A7R III and will presumably do a proper revamp in the next model.

1. Higher overall DXO Mark score.

2. I own both the A7R III and D850. The Nikon has the edge for dynamic range at low ISOs and superior resolution. In video mode, it produces a cleaner 4K full frame image from the sensor, with less moire, due to a superior readout.

The design and manufacturing are both important. Yes, Sony and Samsung have the most cutting edge fabs, but since design is so important, Nikon can still differentiate on performance.

Nikon have a history of their own in-house sensors with dramatically different performance and specs to Sony. The engineers and people are not '10 a penny' they are some of the most respected experts in the industry and very experienced.

Starting with the D1X's CCD sensor way back, which used a Nikon bayer pattern with pixels larger vertically than horizontally.

Then they produced JFET-LBCAST sensor type, different to both CCD and CMOS technology with the D2H.

The D3X sensor outperformed the Sony rival at the time, also a 24MP full frame CMOS, in the A900. I own both the D3X and A900 and can confirm the difference. Mainly noise, especially readout noise.

The D5 uses a radically different sensor than in any Sony camera.

Regarding supposed sales of Nikon-designed sensors to Pentax (D800 36MP) and Fujifilm (D7200 sensor), these were not Nikon sensors. The Nikon sensors were the D500 (20MP APS-C) and D850. Canon also sometimes uses Sony off-the-shelf sensors, such as the 1" 20MP chip in the G7X.

It is clear, reading these responses, that Nikon do need articles like Dave's as part of a PR push, as there are MANY misconceptions out there.

[Robert Collins]

21 minutes ago, Andrew Reid said:

The D850 46MP sensor does outperform the latest Sony 42MP.

1. Higher DXO Mark score.

2. I own both the A7R III and D850. The Nikon has the edge for dynamic range at low ISOs and superior resolution. In video mode, it produces a cleaner 4K full frame image from the sensor, with less moire, due to a superior readout.

The design and manufacturing are both important. Yes, Sony and Samsung have the most cutting edge fabs, but since design is so important, Nikon can still differentiate on performance.

Nikon have a history of their own in-house sensors with dramatically different performance and specs to Sony.

Starting with the D1X's CCD sensor way back, which used a Nikon bayer pattern with pixels larger vertically than horizontally.

Then they produced JFET-LBCAST sensor type, different to both CCD and CMOS technology with the D2H.

The D3X sensor outperformed the Sony rival at the time, also a 24MP full frame CMOS, in the A900. I own both the D3X and A900 and can confirm the difference. Mainly noise, especially readout noise.

The D5 uses a radically different sensor than in any Sony camera.

Regarding supposed sales of Nikon-designed sensors to Pentax (D800 36MP) and Fujifilm (D7200 sensor), these were not Nikon sensors. The Nikon sensors were the D500 (20MP APS-C) and D850. Canon also sometimes uses Sony off-the-shelf sensors, such as the 1" 20MP chip in the G7X.

You are making 2 arguments here....

1) That Nikon is better at 'designing' sensors than say 'Sony'

Although I dont necessarily agree, I will let you have that one. 

2) That the 'design' of sensors is more important than the 'manufacturing'.

I think this is totally wrong.

Let's face it. Is Canon crap at 'designing' sensors or is their manufacturing capacity 'out of date'? Does Canon not have 'stacked cmos' or 'BSI' because they cant 'design' it or cant 'manufacture' it?

Do we know that D850 is a Sony manufactured sensor because it behaves like one or is designed like one?

In the past couldnt you tell whether a Panasonic camera had a 'Sony sensor' or a 'Panasonic sensor' from the performance?

There is a rule of thumb in sensor manufacture that basically says 'each generational leap' requires '2x the capital' which results in 'half the number of players' which resulted in Panasonic selling its ancient manufacturing plants to Tower Jazz, Toshiba to Sony and the fact that Canon cannot afford to move their plants to the latest tech.... So talking about history is dodgy here - perhaps 10 years ago Canon had the best sensors, it doesnt now.

 

[Andrew Reid]

Fair points, but I didn't say design was MORE important - it is critically important, not more or less. You can't have manufacturing without design and visa versa. Especially the kind of design and simulation Nikon revealed in the article. When you are designing at the level of an individual photon and atom, it makes a difference.

Canon's process is less well documented. Over the past 10 years there has been less useful inside information about Canon's sensor design and manufacture capabilities as there was about Nikon in that ONE article. I know Canon recently invested billions in new chip making facilities in Taiwan, but they are not at the Samsung or Sony level, that's for sure.

The machines that MAKE the machines, of course are important and Nikon make that technology to don't forget, the chip lithography side of the business is doing very well.

Until this Imaging Resource article all we had were SONY logos printed on Nikon sensors, pictured on rumours sites so not exactly a world class insight.

We are armed with quite a different set of facts now. And yes, although it is something of a staged PR piece, it is backed by substance - both the facts and evidence are there for all to see.

BSI sensor technology is a capability of the Sony fabs that make Nikon sensors but is not exclusive to Sony, and if you are simulating chip manufacturing at the level of fine detail Nikon is (at atomic level) then you are arguably using a fab in-house, to prototype your own designs, which is almost as good as having your own manufacturing facilities in-house.

Nikon's team clearly are on the cutting edge and differentiating their image quality from Sony. If Sony turn around and decide not to take on contract manufacturing for Nikon sensors any more, there are plenty of other factories... they are 10 a penny Samsung for one.

Canon only just moved to a copper process from aluminum, and it is one of the reasons they have lagged in video on high resolution sensors, as the readout on an aluminum chip is slower - the chip runs hotter, due to the higher resistance of aluminum vs copper. The manufacturing is more difficult and cutting edge behind a copper CMOS circuit, and that in turn was why Samsung were able to do a 6K sensor readout on the NX1 way back in 2014, before Sony and before pretty much everyone.

Next innovation to be widely used is Quad Bayer, I think. It allows 2 exposure sensitives in one pixel, so HDR on chip, essentially. That's design, not just manufacturing. Patents are key and Nikon's team have quite a few.

Canon with Dual Pixel AF have more.

Design is important guys.

[webrunner5]

I think a lot of Nikon's performance is not so much a "different" Sensor, it is the processing After the Sensor that is the difference. They are adding Their Color Science, Their base ISO on and on. Nikon and who ever, even Panasonic is not going to brag about their camera has a Sony sensor in it, they are going to say say it is Their magic sauce, and that is all after the fact of the Sensor. Sony is not making 5 different FF sensors, they are doing 1 or 2 at most. It would cost a Trillion dollars to have all those different Fab lines. Sony is Not going to build a whole new Fab line for what few sensors Nikon is going to buy for a D850, get a grip. It might be in this new Mirrorless one also?? But I would think they have to be different for a DSLR and a Mirrorless. That is the main reason Canon and Nikon don't change their high end stuff only every 3 or 4 years. They milk the Tits off of the Sensor to get their money back to build a new Fab line years later. Hell Canon used that same crappy ass 18mp sensor for like 6 years or more.

Since Sony sells millions of sensors they have the cash to upgrade their Fab line all the time. Money talks. And yeah, Sony MADE the sensor, not Nikon.. Sure there has to be a collaboration between the company's needs, not everyone wants the same things.

 

 

Dave Etchells Mod Didaskalos • 19 hours ago

"I need to preface these comments with the note that I have absolutely no knowledge of any of Nikon's relationships with semiconductor fab providers, so the following is just educated guesses, based on common industry business models...

That said, it's very common in the semiconductor industry for companies to design chips (of all sorts, not just sensors), and then pay "fab" providers to produce them for them. Semiconductor fabrication facilities are insanely expensive, with a billion dollars being pretty much the cost of entry, and costs of $3-4 billion not at all unusual. With that kind of cost, you have to run an enormous volume of manufacturing to get a payback on your capital.

This has led to a business model that often splits the design of chips from their manufacture, with a number of companies offering "fab" services to many different customers. In these case, the customers provide the photolithography masks that define the circuitry on the wafer, through all of the different stages of the fabrication process. (Ion implantation, oxide growth, metallization and insulation layers, etc) The fab company then simply uses the provided masks to run wafers through their factory.

The fact that the D850 sensor has a Sony marking on it suggests that Sony did the manufacturing for it, but that wouldn't be at all inconsistent with Nikon designing the circuitry. I don't know anything about the inner workings of fabs, but it seems like any wafers going through a semiconductor fab would need to have an identifying code etched into them, as a fail-safe identification method that would distinguish one batch of wafers from others within the facility.)

Again, Nikon didn't share any information with me about who they use for fabrication, so I have no direct knowledge of any business relationships involved, but given the markings on the chip, it seems likely that Sony provided fabrication services for the D850 sensor, which Nikon designed".

[Robert Collins]

12 minutes ago, Andrew Reid said:

Nikon's team clearly are on the cutting edge and differentiating their image quality from Sony. If Sony turn around and decide not to take on contract manufacturing for Nikon sensors any more, there are plenty of other factories... they are 10 a penny Samsung for one.

It really comes down to this doesnt it?

But you have to ask why?

Nikon uses Sony sensors

Olympus uses Sony sensors

Panasonic uses Sony sensors

Fuji uses Sony Sensors

Pentax uses Sony sensors

Hasselblad uses Sony sensors

Phase One uses Sony sensors

Blackmagic uses Sony sensors

And even Samsung, Apple and Canon use Sony sensors in many of their products. 

It isnt because CMOS sensor manufacturers are ten a penny but camera manufacturers......

[MdB]

1 hour ago, Andrew Reid said:

The D850 46MP sensor does outperform the latest Sony 42MP. It is a newer design, Sony has tweaked an old design for the latest A7R III and will presumably do a proper revamp in the next model.

1. Higher overall DXO Mark score.

2. I own both the A7R III and D850. The Nikon has the edge for dynamic range at low ISOs and superior resolution. In video mode, it produces a cleaner 4K full frame image from the sensor, with less moire, due to a superior readout.

You are choosing 'some' areas and stating it is unequivocally 'better'. Better at some things yes. One would expect it to be given it is the latest in what Sony can make vs a 2-3 year old design. This is my bet: 

Sony 42MP is exclusive to Sony. It is Sony designed for Sony cameras. Specifically around the on sensor PDAF. 

The D850's 46MP sensor is also a Sony design. It is the Sony third party sensor replacement for the aging 36MP model (sold to Nikon and Pentax). The 46MP, like the 36MP before it has an exclusivity arrangement with Nikon for a period of time. After that time we will see a K-1 III rocking that sensor most likely. Nikon toppings and packaging notwithstanding. 

1 hour ago, Andrew Reid said:

Nikon have a history of their own in-house sensors with dramatically different performance and specs to Sony.

Sure. As I said they made their own sensors. They also got other manufacturers to make their sensors (like the one in the D4 / Df). Not sure what that has to do with Sony sensors? 

1 hour ago, Andrew Reid said:

The D5 uses a radically different sensor than in any Sony camera.

And it has disappointing performance to say the least. 

 

1 hour ago, Andrew Reid said:

Regarding supposed sales of Nikon-designed sensors to Pentax (D800 36MP) and Fujifilm (D7200 sensor), these were not Nikon sensors. The Nikon sensors were the D500 (20MP APS-C) and D850. Canon also sometimes uses Sony off-the-shelf sensors, such as the 1" 20MP chip in the G7X.

So Nikon have these designers that are the best in the world and make way better designs than Sony can, yet still use Sony designs? Seems somewhat contradictory. D800 was a 'Nikon' sensor until it wasn't. So was the D7200. Yes Canon and everyone else uses off the shelf Sony sensors even though they have their own designers too... Wonder why Canon don't just get Sony to make their designs? 

1 hour ago, Andrew Reid said:

It is clear, reading these responses, that Nikon do need articles like Dave's as part of a PR push, as there are MANY misconceptions out there.

It's clear reading this article that people will believe anything that aligns with their own beliefs. 

57 minutes ago, Andrew Reid said:

Fair points, but I didn't say design was MORE important - it is critically important, not more or less. You can't have manufacturing without design and visa versa. Especially the kind of design and simulation Nikon revealed in the article. When you are designing at the level of an individual photon and atom, it makes a difference.

Honestly I haven't read anything that says that they are 'designing' at that level. They are testing reference designs and making design tweaks to packaging and toppings. 

59 minutes ago, Andrew Reid said:

Until this Imaging Resource article all we had were SONY logos printed on Nikon sensors, pictured on rumours sites so not exactly a world class insight.

Sony have spoken about the relationship they have with their customers pretty frankly in the past. I seem to recall a DPR article where Sony were talking about the process for (say) the D850 sensor... Yet Team Nikon Boys still didn't think the D850 was even fabbed by Sony (until very recently when it was confirmed). Nikon Crew simply don't like Sony having any part of their camera. 

1 hour ago, Andrew Reid said:

We are armed with quite a different set of facts now. And yes, although it is something of a staged PR piece, it is backed by substance - both the facts and evidence are there for all to see.

I'm still struggling to see what the 'facts' being presented are. Fact 1: Nikon design packaging. Fact 2: Nikon design toppings. Fact 3: Sony seem to do the rest. This has pretty much always been known. Fact 4: Sony don't use these sensors themselves except in some rare cases and as such most are not directly comparable. 

1 hour ago, Andrew Reid said:

BSI sensor technology is a capability of the Sony fabs that make Nikon sensors but is not exclusive to Sony, and if you are simulating chip manufacturing at the level of fine detail Nikon is (at atomic level) then you are arguably using a fab in-house, to prototype your own designs, which is almost as good as having your own manufacturing facilities in-house.

Say what? 

1 hour ago, Andrew Reid said:

Nikon's team clearly are on the cutting edge and differentiating their image quality from Sony. If Sony turn around and decide not to take on contract manufacturing for Nikon sensors any more, there are plenty of other factories... they are 10 a penny Samsung for one.

Interesting that Nikon take their manufacturing to Sony then given how much they go about hiding the fact. Why aren't one of those other makers making their sensors? Seems a bit weird for Nikon to hand over their masterful 'designs' to the competition. 

1 hour ago, Andrew Reid said:

Next innovation to be widely used is Quad Bayer, I think. It allows 2 exposure sensitives in one pixel, so HDR on chip, essentially. That's design, not just manufacturing. Patents are key and Nikon's team have quite a few.

Patents are a whole different discussion. Quad Bayer Sony just announced. You would think they'd have to wait for all the designers to announce theirs first  

 

[Andrew Reid]

2 hours ago, MdB said:

The D850's 46MP sensor is also a Sony design. It is the Sony third party sensor replacement for the aging 36MP model (sold to Nikon and Pentax). The 46MP, like the 36MP before it has an exclusivity arrangement with Nikon for a period of time. After that time we will see a K-1 III rocking that sensor most likely.

You still did not read either article. So very, very annoying.

3 hours ago, Robert Collins said:

Nikon uses Sony sensors

Not in the flagship two models. Only in the lower-end cameras.

Quote

Olympus uses Sony sensors

They've also used Panasonic sensors. And so what?

iPhones are designed by Apple, with custom silicon involving ARM and in-house Apple chip designers, but it ends up being made in a Chinese factory owned by Foxconn.

Does that make it the Foxconn iPhone?

That's your logic here Rob!

It does not come down to your simplification at all.

If you define a "Sony sensor" as being any sensor manufactured in a Sony fab, then that is just over-simplification and wrong.

Quote

Panasonic uses Sony sensors

Yes they do. They also continue to develop their own next-generation organic sensor with Fujifilm.

Quote

Fuji uses Sony Sensors

Again like Nikon they have custom designs and custom image pipeline, from readout to bespoke image processor.

You think X-Trans is a Sony technology?

Quote

Pentax uses Sony sensors

Pentax does take off-the-shelf parts, yes.

Quote

Hasselblad uses Sony sensors

Phase One uses Sony sensors

Only the CMOS models.

The CCD sensors were not Sony.

Quote

Blackmagic uses Sony sensors

Yes but they have also used Fairchild, CMOSIS, all sorts.

Quote

And even Samsung, Apple and Canon use Sony sensors in many of their products. 

It isnt because CMOS sensor manufacturers are ten a penny but camera manufacturers......

To be honest Rob I still don't see how your point applies to Nikon designing their own sensors.

You've mentioned a lot of other manufacturers. So what?

[keessie65]

What about the Samsung / Fuji sensor? I think Nikon is blowing their sensor away. Maybe only Fuji uses the chip, but can he compete with the new Nikon? I am more and more of the opinion that Samsung has nothing to say in terms of cameras and sensors, unfortunately ...
However, I still hope for Samsung's comeback. But lose confidence that they will actually do this

[MdB]

6 hours ago, Andrew Reid said:

You still did not read either article. So very, very annoying.

Actually I did. Did you really read it? Based on what you've said you have simply gone with the dominant reading of the article. This is what Nikon want, so well done there. It's a nice little propaganda piece put together for the launch of their new mirrorless system and so we'll quietly forget they just abandoned another mount. 

Didn't bother to answer any of the discussion points? Very, very annoying. I did read the article and see nothing that suggests that Nikon design anything beyond packaging and sensor toppings, which is exactly what we've already always known. There is a LOT of implication that specific technologies used in the sensors Nikon use came from Nikon, but nowhere does it explicitly state it. 

6 hours ago, Andrew Reid said:

Not in the flagship two models. Only in the lower-end cameras.

Wait, the article said they had whole teams dedicated to every sensor. Why didn't they say only for these models? Why didn't they say only 2 teams? Why didn't they say only for top 'flagship' models? Again, the article doesn't actually say what you think it says. You are taking the implications that are applied by the Nikon PR department (who had control over the whole situation as stated in the article) as being the 'truth'. 

6 hours ago, Andrew Reid said:

iPhones are designed by Apple, with custom silicon involving ARM and in-house Apple chip designers, but it ends up being made in a Chinese factory owned by Foxconn.

This is a silly argument. NOBODY is saying a Nikon camera is a Sony camera because it uses a Sony sensor. Apple also use Sony sensors, nobody is implying that iPhones are Sony phones. But people will say the LCD is an LG or Samsung or whoever. Just as they say the camera sensor is a Sony. Equally, if Apple customise (in whatever way) and ARM processor to their own specs, is it no longer an ARM processor? 

7 hours ago, Andrew Reid said:

Again like Nikon they have custom designs and custom image pipeline, from readout to bespoke image processor.

You think X-Trans is a Sony technology?

Image processors are not image sensors. Again nobody is claiming that Nikon (or anyone else) use Sony image processors. 

As for X-Trans - I would hardly call it a 'technology' in of itself, it is a rearrangement of the colour filter array. Bayer arrays are not Sony's (on Nikon's) design either. Additionally the CFA is part of the sensor toppings, which Nikon (and Fuji etc) do customise. It is really the main thing that anyone customises. So you are pointing again at the same thing and saying 'see, see?'. 

7 hours ago, Andrew Reid said:

Pentax does take off-the-shelf parts, yes.

Oh of course, Pentax are too weak and feeble and unlike the mighty Nikon they take off the shelf sensors. But Nikon 'design' theirs. Why is it that the Pentax off the shelf sensors seem to be a LOT like the Nikon sensors? 

7 hours ago, Andrew Reid said:

To be honest Rob I still don't see how your point applies to Nikon designing their own sensors.

Let me ask you this: If Nikon are responsible for the design of these sensors and the fabbing plant literally just makes them to Nikon spec, why don't Nikon diversify their production? I mean we've had earthquakes and other disasters that have left Nikon waiting and waiting for sensors, why not have multiple fabbing companies just make them for them? They are having supply issues with the D850, why not get Reneasis to make some D850 sensors as well, keep the supply chain going. Apple diversify their parts so that if a supplier goes down or has QC issues or whatever they can maintain production. Why don't Nikon for each sensor? 

[ND64]

D850 pixel stores 10,000 more electrons than A7R3 pixel. So ISO 64 is not just a gimmick number that anybody can add to the spec list. Its indication of a state of the art read out circuit that is capable of handling such a massive amount of electrons. Also, Sony's advantage is not their fab devices and tools (they're far behind Samsung, TSMS, GloFo in nod size anyway), their advantage is their huge asset of IP. AND some of those IPs are bought from Nikon, like PDAF and some Exmor goodies. 

[Andrew Reid]

The facts are pretty clear to me! Fact is Nikon can design their OWN sensor independent of Sony. Toshiba were making Nikon's custom D5300 sensor at an independent fab before being bought by Sony, and this was another design that outperformed what Sony was doing at the time. Nikon can go to Samsung and have a custom sensor mass-produced that is unique on the market. They don't absolutely need Sony and it is not a case of just packaging up a Sony sensor, tweaking this or that...like @MdB is claiming.

So you have to give Nikon much more credit...

That they don't own the building where the mass-production happens, doesn't mean anything.

It's the same with ARM... And nobody denies there are ARM chips in our smartphones.

9 hours ago, MdB said:

Oh of course, Pentax are too weak and feeble and unlike the mighty Nikon they take off the shelf sensors. But Nikon 'design' theirs. Why is it that the Pentax off the shelf sensors seem to be a LOT like the Nikon sensors?

Because there are established design rules and pixel architectures that everybody uses because they work well.

I wouldn't say the 36MP K1 II chipset is anything like a Nikon D850 + Expeed so I think the premise of your point is kinda flawed.

Unless the K1 II suddenly gains 10 megapixel and starts outputting 4K video, it's not "A LOT like a Nikon sensor"

9 hours ago, MdB said:

As for X-Trans - I would hardly call it a 'technology' in of itself, it is a rearrangement of the colour filter array. Bayer arrays are not Sony's (on Nikon's) design either. Additionally the CFA is part of the sensor toppings, which Nikon (and Fuji etc) do customise.

Apart from talking about a complex sensor like a cake, do you really understand the implications of changing such a fundamental sensor architecture? It's not a topping like some butter spread on a pancake. It is an architecture which has a knock-on effect on the de-bayering, image processing, micro lenses, and a whole lot more. So all that needs to be customized as well.

CMOS sensors are fundamentally printed circuits and the machines in Sony's factories use photolithography to produce a wafer of chips https://en.wikipedia.org/wiki/Photolithography

Guess who is one of the dominant supplier of these machines... Yes, Nikon!

As far back as 1984 they had a large chunk of the semiconductor lithography market.

So not only do they design their own camera CMOS sensors, they make the machines for Sony's factory and others.

So a custom design, printed by Nikon's own machines... And yet some people still can't except that it is a Nikon sensor, must be a Sony off-the-shelf cake with a Nikon topping of icing?!

Currently market share of Nikon in lithography is 20%

[MdB]

2 hours ago, Eric Calabros said:

D850 pixel stores 10,000 more electrons than A7R3 pixel. So ISO 64 is not just a gimmick number that anybody can add to the spec list. Its indication of a state of the art read out circuit that is capable of handling such a massive amount of electrons.

This comes at the expensive of high ISO noise. For some people one is an advantage and for others the other. That doesn't make a single clear cut 'better' sensor. 

 

2 hours ago, Eric Calabros said:

AND some of those IPs are bought from Nikon, like PDAF

Actually Aptina. 

1 hour ago, Andrew Reid said:

Toshiba were making Nikon's custom D5300 sensor at an independent fab before being bought by Sony

So wait, Nikon ONLY make custom sensors in their flagship cameras (your words) AND the budget D5300? What an absolute load of BS. Toshiba was making a better sensor than Sony at the time? Probably. Not really the topic of discussion. If I recall that sensor had a lot of banding issues in shadows. 

 

1 hour ago, Andrew Reid said:

So you have to give Nikon much more credit...

Sorry because you say so? Still haven't seen any evidence otherwise. 

1 hour ago, Andrew Reid said:

It's the same with ARM... And nobody denies there are ARM chips in our smartphones.

Did you just agree with me? Perhaps you didn't understand my point. 

1 hour ago, Andrew Reid said:

Because there are established design rules and pixel architectures that everybody uses because they work well.

What?! Oh there was a magic book that fell to earth with the 'standard' designs... what rot. 

1 hour ago, Andrew Reid said:

wouldn't say the 36MP K1 II chipset is anything like a Nikon D850 + Expeed so I think the premise of your point is kinda flawed.

Who said the D850?! The K1 uses a Sony made sensor of 36MP... sound familiar? Sounds an awful lot like the ones in the D800 / D800e / D810 / A7R. These were 'flagship' models and the claim here is that Nikon have been doing this for 'years'. Or did they start with the D850, because it's convenient for the arguement (as nobody else uses that sensor yet)? 

1 hour ago, Andrew Reid said:

Apart from talking about a complex sensor like a cake, do you really understand the implications of changing such a fundamental sensor architecture? 

Do you? I didn't know you were a leading expert on sensor designs. The fact that you seem confused by this seems to suggest otherwise. YES toppings on sensors are a lot like cake. They even talk about these things in the article you supposedly fully absorbed. Real camera companies like Phase One don't pretend to design these sensors themselves and discuss things with far greater openness, depth and understanding than the PR stunt. Maybe read what those companies have to say on the matter.

Custom colour filter arrays are completely normal. They don't always change the layout, but they do change the filtering etc. Fuji changing to X-trans array does have processing implications for them, but that has zero to do with the sensor. Fuji still didn't design it.

1 hour ago, Andrew Reid said:

It is an architecture which has a knock-on effect on the de-bayering, image processing, micro lenses, and a whole lot more. So all that needs to be customized as well.

Micro lenses and CFA are essentially the customised toppings. Debayering IS image processing and has nothing to do with sensors. And again we are talking about the couple things these manufacturer have some input into and making it seem like they designed the whole shebang. 

1 hour ago, Andrew Reid said:

So a custom design, printed by Nikon's own machines... And yet some people still can't except that it is a Nikon sensor, must be a Sony off-the-shelf cake with a Nikon topping of icing?!

 

Sounds like Sony aren't a fabbing company, they are a warehousing company for Nikon, who clearly are the leading sensor design and manufacturers, Sony are a glorified shed. Clearly Sony have no clue what they are doing because some  fanboys want to believe.

1 hour ago, Andrew Reid said:

Currently market share of Nikon in lithography is 20%

Which is only a fraction of the market Sony has in sensors. But apparently they're all Nikon sensors made on Nikon machines by Nikon engineers with profits going to Nikon and all IP owned by Nikon because they are clearly the bestest at everything. Sony are just fortunate enough to let them store them in their empty shed because they don't have anything else to do. 

 

Couldn't answer one question. Again.