Pages: 1 ... 18 19 [20] 21 22 ... 29   Go Down

Author Topic: Nikon D850: Announcement of an Announcement of Development  (Read 929543 times)

scyth

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #380 on: August 27, 2017, 10:52:47 pm »

BSI would deliver that
that (performance @ "high ISO") was delivered by DR-Pix not BSI ...
Logged

hogloff

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 656
Re: Nikon D850: a good custom-designed sensor
« Reply #381 on: August 27, 2017, 10:57:34 pm »

market share dynamics so far proves Canon right...

Yep...wonder what it costs Nikon to have their sensor manufacturing outsourced? I agree...Canon seems to continue to run well inspite of themselves. They've been gaining market share at Nikon's expense.
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: a good custom-designed sensor
« Reply #382 on: August 27, 2017, 11:02:17 pm »

market share dynamics so far proves Canon right...

Correlation does not imply causality. That Canon holds a market share advantage does not imply that it is due to keeping things in-house.

Canon got a huge head start in digital with CMOS and full-frame, then with the 5D and 5D2, causing people to buy into the system. This led to Canon taking the lion's share of the high-end market in the early days; this has slowly bled away as Nikon caught up and, in some areas, overtook Canon, but, due to the shelf life of lenses and the expense of changing systems, still hold the advantage (if not as large as it was previously).

It could just as easily have gone the other way - had Nikon been first off the ground with full-frame, CMOS and a camera competitive with MF film, they might be the market leqders now (and resting on their laurels) instead of Canon.
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #383 on: August 27, 2017, 11:05:36 pm »

that (performance @ "high ISO") was delivered by DR-Pix not BSI ...

They're not mutually exclusive. Sony uses BSI (and is currently the only company to use it). It's unlikely they won't have used both in this sensor.
Logged

hogloff

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 656
Re: Nikon D850: a good custom-designed sensor
« Reply #384 on: August 27, 2017, 11:07:02 pm »

Correlation does not imply causality. That Canon holds a market share advantage does not imply that it is due to keeping things in-house.

Canon got a huge head start in digital with CMOS and full-frame, then with the 5D and 5D2, causing people to buy into the system. This led to Canon taking the lion's share of the high-end market in the early days; this has slowly bled away as Nikon caught up and, in some areas, overtook Canon, but, due to the shelf life of lenses and the expense of changing systems, still hold the advantage (if not as large as it was previously).

It could just as easily have gone the other way - had Nikon been first off the ground with full-frame, CMOS and a camera competitive with MF film, they might be the market leqders now (and resting on their laurels) instead of Canon.

Where are you getting your figures from. Nikon's market share has been going down for years now...sitting currently at less than 25%. Canon's share has stayed steady at around 50%. be interesting to see if this changes...but even Nikon's projections do not indicate much change to this trend.
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: a good custom-designed sensor
« Reply #385 on: August 27, 2017, 11:26:55 pm »

Where are you getting your figures from. Nikon's market share has been going down for years now...sitting currently at less than 25%. Canon's share has stayed steady at around 50%. be interesting to see if this changes...but even Nikon's projections do not indicate much change to this trend.

25% by what? Units sold (making no distinction between high end and low end)? Gross sales? Net profit on camera bodies? Net profit on bodies and lenses? You'll get different market share figures depending on which measure you look at. It's common knowledge that Canon had flooded the low end with Rebels and xxD bodies, while Nikon has been improving the top end. But the low end is a dying sector.
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Online Online
  • Posts: 2162
    • The Last Word
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #386 on: August 27, 2017, 11:30:02 pm »

just don't forget that Ph2Ph does not calculate per sensel PDR... so it is the same PDR assuming the ideal downsizing... and raw converters/post processing are not ideal math-wise once you start doing demosaicking and color transforms first.

Bill and I don't assume ideal downsizing for our PDR normalization. We assume standard downsizing. In tests that I've done, raw converters can do better than that by using smart noise reduction at high res.

Jim

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: a good custom-designed sensor
« Reply #387 on: August 28, 2017, 12:10:31 am »

I continue to strongly dispute the proposition that it would be in the best interest of Sony not to sell their best sensors to Nikon. Very few of the sensors they would not sell in Nikon bodies would end up in Sony bodies. The cross selling btwn these 2 is much more limited than they seem to think. Sony would end up both losing an important income for their sensor division and also indirectly help strengthen the competitor Nikon would end up working with.

There's a big difference between making the best sensor Nikon can design for them, and selling them the best Sony-designed sensor, even before it makes it into a Sony camera. The former makes sense - if Sony didn't make it for them, someone else would, and they'd lose business. The latter would be a good way to squander your lead in sensor design and kill off your own camera division. They don't even sell their 42MP sensor at the moment (unlike the 36MP, it's not listed on their website as a commercially-available sensor). Companies which do not make cameras can sell off their best design; companies which do cannot afford to do so, unless they don't intend to compete in that sector anyway. Which probably explains why the D850 is a Nikon design made by Sony, rather than a pure Sony design.

Quote
The D850 sensor should indeed clearly put an end to this story... but it won't... because now they are coming up with the theory that 46mp isn't a high resolution sensor... and that the real best sensor of Sony will end up in these... ;)

It's not high-resolution because it's capable of 9fps.

If you can drive a 46MP sensor at 9fps, you can drive an 80MP sensor at 5fps. And Sony would have no difficulty making an 80MP sensor, since they make phone sensors which are even denser, and would have no problem making one for Nikon if Nikon had designed and requested one.

Clearly, the decision to make it 46MP rather than something higher was part of a tradeoff to allow it to shoot at 9fps (whether this tradeoff was voluntary or forced is another matter of little relevance here). If Nikon had wanted a resolution-focused design, they'd have gone with a high-resolution sensor with a slower frame rate. Certainly, Sony and Canon are both likely to do this with their 5Ds and A7r2 replacements (although I hope they also offer a mid-resolution, mid-high frame rate design like the D850). But, instead, they went for speed, and, as a result, used a sensor with a resolution that, while high compared with current cameras, is not as high as could otherwise have been used had they not also gone for 9fps.
Logged

henrikfoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 899
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #388 on: August 28, 2017, 03:37:08 am »

With Canon at 50mp and Nikon at 45mp I think the new Sony r camera will make a real jump.
I think we can expect more than 60mp.
Logged

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #389 on: August 28, 2017, 03:44:52 am »

With Canon at 50mp and Nikon at 45mp I think the new Sony r camera will make a real jump.
I think we can expect more than 60mp.

Same with the 5D2's replacement.

Canon were at 50MP two years ago. They're not going to go backwards. And now they've got on-sensor A/D conversion, which will improve DR at the same time.

Only thing is, with a 60MP-plus sensor, neither the Canon or Sony will be fast. They will not be competitors with the D850, any more than the D5 or A9 are. Canon and Sony may release other models which will compete in the same space, but they are unlikely to be their highest-resolution studio/landscape cameras.
Logged

BernardLanguillier

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 12536
    • http://www.flickr.com/photos/bernardlanguillier/sets/
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #390 on: August 28, 2017, 03:55:25 am »

With Canon at 50mp and Nikon at 45mp I think the new Sony r camera will make a real jump.
I think we can expect more than 60mp.

Rumors were saying 70mp.

Now, the relevance of this needs to be assessed. There are very few wides that are able to resolve even 46mp well in my book. Even what may be the best wide available, the Nikon 19mm T/S may be challenged at that level of resolution. The situation of course differs for the latest tele lenses that have plenty of resolution.

As a result, for wide applications, stitching becomes a much better way to reach significant increases of real detail.

Another option is of course small MF such as the Fuji 23mm that will be fine at 100mp.

Cheers,
Bernard

shadowblade

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2605
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #391 on: August 28, 2017, 04:28:43 am »

Rumors were saying 70mp.

Now, the relevance of this needs to be assessed. There are very few wides that are able to resolve even 46mp well in my book. Even what may be the best wide available, the Nikon 19mm T/S may be challenged at that level of resolution. The situation of course differs for the latest tele lenses that have plenty of resolution.

As a result, for wide applications, stitching becomes a much better way to reach significant increases of real detail.

Another option is of course small MF such as the Fuji 23mm that will be fine at 100mp.

Cheers,
Bernard

Depends what you're trying to shoot and whether you care more about centre or corner sharpness, and wide-open or stopped down. Certainly, they can all handle it in the centre. The Canon 16-35 can also handle the corners at 50MP on the 5Ds, while the new Sony 12-24 and 16-35 can do so at 42MP on the A7r2. No doubt an updated Nikon 14-24 could do thr same at 46MP. Stopped down, they could probably make use of 60-80MP in the corners, and, when you're shooting a scene that needs to be sharp corner-to-corner, you're probably shooting stopped down.

The Nikon lens I'd like to see most for the D850 is actually an updated 200-400 f/4, with an inbuilt TC. Possibly even a 200-500 f/4, if they can do it without making it too heavy. It's overdue for an update and would show off the D850's unique capabilities like no other lens could.

Do you mean using a MF camera for wide shooting, or attaching a MF lens to a FF body? If it's the latter, MF lenses aren't actually all that sharp - less sharp than their FF equivalents. Just that they don't need to be as sharp, since they're normally being used on larger sensors with larger photosites.
Logged

hogloff

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 656
Re: Nikon D850: a good custom-designed sensor
« Reply #392 on: August 28, 2017, 11:15:46 am »

25% by what? Units sold (making no distinction between high end and low end)? Gross sales? Net profit on camera bodies? Net profit on bodies and lenses? You'll get different market share figures depending on which measure you look at. It's common knowledge that Canon had flooded the low end with Rebels and xxD bodies, while Nikon has been improving the top end. But the low end is a dying sector.

Overall revenue ( $$$ ).
Logged

BJL

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 6418
Re: Nikon D850: a good custom-designed sensor
« Reply #393 on: August 28, 2017, 09:19:46 pm »

... I think we can can stop worrying that Nikon is hampered by having to settle fpr whatever second-best sensors that Sony puts in its catalog. Maybe at times Canon hampers itself by insisting too much on doing everything in-house,  it of course it could always drop that if it fell too far behind in sensor fab ability: Sony would love a huge customer like Canon.
In case anyone misunderstood, I am not coming down for either "in-house" or "outsource" as always being the best way. (It does seem to me that designing at least partially in-house in order to have unique offerings, and then out-sourcing fabrication wen that reduces costs, is often a good intermediate strategy.)

In-house might be most profitable in some cases, by adding profits to the company's manufacturing branch, but sometimes manufacturing costs are significantly lower for a very large, efficient out-sourcing fab. A fascinating case is Samsung: a huge electronics company that designs and produces Exynos processors for some of its phones, and yet outsources some processors from Qualcomm for other models. Also look at what has happened to CPUs for desktops, laptops and servers: outsourcing from Intel has become the dominant strategy.

And as others a have said, many factors can contribute to Canon's market success, so one cannot conclude that it is due specifically to its in-house approach to DSLR sensors. This only shows that designing and producing DLSR sensors entirely in-house is _one_ viable option.
Logged

scyth

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 584
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #394 on: August 28, 2017, 09:57:51 pm »

Bill and I don't assume ideal downsizing for our PDR normalization. We assume standard downsizing.

OK

1) what is standard downsizing ?
2) what kind of demosaick and color transforms you apply before downsizing ?
Logged

Jim Kasson

  • Sr. Member
  • ****
  • Online Online
  • Posts: 2162
    • The Last Word
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #395 on: August 28, 2017, 11:56:49 pm »

OK

1) what is standard downsizing ?
2) what kind of demosaick and color transforms you apply before downsizing ?

1) The more-or-less standard linear methods. Read this and the following posts to get a sense of the relationship between noise and detail with linear downsampling (there is a certain amount of devil in the details):

http://blog.kasson.com/the-last-word/noise-reduction-and-downsampling/

2) Here's an example of the advantages of higher resolution sampling, and nonlinear noise reduction:

http://blog.kasson.com/the-last-word/noise-reduction-with-nonlinear-tools-and-downsampling/

And let me agree that the noise behavior that Bill and I assume is in a sense "ideal" for linear filtering, but you can do a lot better with nonlinear filtering followed by downsampling.

Jim
« Last Edit: August 29, 2017, 12:06:59 am by Jim Kasson »
Logged

davidgp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 757
    • davidgp fotografia
Re: Nikon D850: a good custom-designed sensor
« Reply #396 on: August 30, 2017, 04:50:07 am »


And, once again, these Nikon naysayers are very quickly forgetting that the Nikon orders pretty funded for many many years the development of the great sensor technology we see today in APS-C and FX bodies. I am fully aware that mobile phones were another key funding stream, but there is a difference between a mobile phone sensor and one for a DSLR.


Sorry for the late reply to this comment. I just wanted to clarify a point, there is no difference between making a mobile phone sensor or a full-frame one, it is exactly the same process. That it is due to the CMOS technology, the same one used to make nearly any chip inside any device, from cameras, to computers, to microwaves... well, there is one key difference, it is much cheaper to try new things in smaller sensors than big ones.

All CMOS technology it is build nowadays in wafers of pure Silicon (as pure as possible, any impurities can make a batch of sensors just a paperweight) of 300mm of diameter (some years ago they were using 200mm, and there was speculation of moving to 400mm for some years, but never happen). Each of these wafers can be used to build several sensors (or processors, or RAM memory...), if they chip chip is small, you can put more chips in the same wafer. This is done by a series of physical and chemical process... very precise one, since we are talking of few nanometres here for the precision of the process. Doing a quick search in Google, you can see here a wafer with several 35mm sensor chips: https://www.google.com/search?q=sony+sensor+wafer&rlz=1C1GGRV_esES752__752&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjRm5_8xf7VAhUQsBQKHQhBCpYQ_AUICygC&biw=1920&bih=990#imgrc=POlZJr58ajWwhM:

Any impurity in the silicon in one of those chips, sensor to the waste, any error in the physical or chemical process for just impurity in the air, or the chemical, or whatever... chip to the waste... It is expected that not all the chips in a wafer work... the chip manufactures use what they call a yield rate, number of good chips / number of total chips in the wafer. They try to achive as close to 1 as possible. At the beginning of any technology introduction (new physical/chemical process... etc...) or new chip design, the yield rate tends to be low, it gets higher with the pass of time.

A way that the manufacturers have to get higher yield rates while trying new things, it is to do it with smaller chips, with smaller chips you have more chips per wafer, so even if 40% of the wafer produces bad chips, you still are getting enough chips to sell and compensate the costs... with bigger chips, maybe the majority are bad... and the good chips will be quite expensive. After fine tunning the process with smaller chips, you can feel confidente that you can scale up the technology to bigger chips.

New technologies that are comming to sensors in the lastest years, like Aptina-like amplifiers, BSI (both in the 850, A7r II), or stacked sensor (A9), first released in smartphones.

Here you have a nice paper of evolution of sensor technology: https://semiengineering.com/cmos-image-sensors-cis-past-present-future/

kers

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2256
    • Pieter Kers
Re: Nikon D850: a good custom-designed sensor
« Reply #397 on: August 30, 2017, 05:46:29 am »

...
Here you have a nice paper of evolution of sensor technology: https://semiengineering.com/cmos-image-sensors-cis-past-present-future/

Thank you David - very informative
PK
Logged

henrikfoto

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 899
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #398 on: August 30, 2017, 06:04:19 am »

"New technologies that are comming to sensors in the lastest years, like Aptina-like amplifiers, BSI (both in the 850, A7r II), or stacked sensor (A9), first released in smartphones."

David!  Sorry for my unknowingness, but could you explain what a stacked sensor is?

Henrik
Logged

davidgp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 757
    • davidgp fotografia
Re: Nikon D850: Announcement of an Announcement of Development
« Reply #399 on: August 30, 2017, 07:47:39 am »

"New technologies that are comming to sensors in the lastest years, like Aptina-like amplifiers, BSI (both in the 850, A7r II), or stacked sensor (A9), first released in smartphones."

David!  Sorry for my unknowingness, but could you explain what a stacked sensor is?

Henrik

Here it explains it quite well: https://www.dpreview.com/news/5696183465/sony-shows-off-3-layer-stacked-smartphone-sensor-that-can-shoot-1000-fps-at-1080p

Basically they make a sandwitch: 1st layer: CMOS sensor (it could be BSI or normal one), 2nd layer DRAM, 3rd layer Processor logic. The idea is basically to copy quite fast all the data captured by the sensor to the memory. This makes reading the image quite fast, or much faster than a not stacked sensor. It is ideal for fast framerate (the A9 gets 20 fps full frame at 24 megapixels), also for video will be interesting, with faster reading, less rolling shutter.

Not sure about this, but since you are putting three components together: Sensor+memory+image processor, maybe noise due to heat could increase comparing with a normal sensor (again not sure)... but in this kind of technology it is always a trade off, depending in what the manufacturer wants to prioritize.
Pages: 1 ... 18 19 [20] 21 22 ... 29   Go Up