Nikon’s new mirrorless system, coming in ... late September 2018

[davidgp]

This reminds me of my 27 year career at Intel.  There was absolutely zero cross-talk between the unit that worked with the Windows PC industry and the unit that worked with Apple.  Even when Apple announced it was switching from Motorola to Intel processors, it took virtually everyone, except the super secret Intel/Apple team by surprise.  Even the most senior people, except those that had an absolute need to know, had no idea.

Yes, there is lot of examples like that in tech semiconductor industry...

[chez]

Because Sony Imaging Products commented that several times in interviews... they don't get inside information about what the company is doing to other competitors.

One of the reasons that Sony did making Sony Semiconductors an independent company was to make imaging products competitors (from mobile phones to cameras) less worried that Sony could get inside information from their designs.

There has also been statements that the semiconductor group reserves their latest sensor developments for the Sony cameras...before offering the technology to others.

[davidgp]

There has also been statements that the semiconductor group reserves their latest sensor developments for the Sony cameras...before offering the technology to others.

That was a confusion created by an interview done by Imaging Resource, they clarified that Sony does not keep sensor technology for themselves, you can read the clarification here: https://www.imaging-resource.com/news/2017/03/26/sony-thailand-factory-tour-qa-mapping-out-the-future-of-the-interchangeable together with a very interesting interview.

If you like sensor technology, Imaging Resource has a nice piece also of the process how Nikon designs their sensors using Sony foundries (well, or whatever other foundry they use in the past or future) https://www.imaging-resource.com/news/2018/07/17/pixels-for-geeks-a-peek-inside-nikons-super-secret-sensor-design-lab




http://dgpfotografia.com

[shadowblade]

No:
- I am mentioning 2 cameras
- I am not speaking about price

Cheers,
Bernard

You seemed to be talking about one camera that did everything, covering the A7III, A7r3 and A9.

If it's two cameras, it will happen, but not in the first generation of Nikons - more likely in the next generation of Sonys (A7r4 or A9r).

Matching the A7III is easy enough, although building up a market share from scratch against an established lineup of E-mount lenses may be a bit harder (probably easier for an entry-level body, whose users will often only get two or three lenses at most anyway - as long as you have the bases covered, you're fine).

But there is no way they will match both the A9 and A7r3 in the same camera this year. It's one or the other. The A9 uses a stacked sensor to achieve its AF performance. The A7r3 sensor lacks that, but has close to twice the resolution and higher dynamic range. Same with the D850 sensor (which doesn't even have on-sensor AF). Basing it on the D5 sensor wouldn't help either - not only does it lack the resolution to compete with the A7r3, but it also lacks on-sensor AF, the D5's AF performance being derived from off-sensor components rather than the sensor itself. So, they'd need an entirely new sensor that not only matches the D850 in resolution, but also somehow has an on-sensor AF system that matches the A9 or D5, with the off-sensor data bandwidth and processing power to support 20fps at 45MP. If Nikon could do that, they'd have done it two years ago. More likely, it's something for the A7r4 or A9r (with the next-generation A9 improving the AF further and likely moving to something like 6k or 8k video output, as the 'action' body in the lineup).

[shadowblade]

So, my bet, Nikon will say that both their cameras have the best AF system in their class (Sony always says that in their marketing material... ), fanboys will claim that it is the truth... Anyway, in terms of market, and according to rumors, Nikon will release two cameras:

- A 24 megapixel camera, focusing in prize, that will take the place of their D6XX and D7XX line. This will be a camera with better AF system of the two (Sony A7 III has better AF that the Sony A7r III, in fact, it has the same AF system as the A9, just slower chip reading the image from the sensor, making it imposible to reach 20 fps and that good nearly no-rolling-shutter electronic shutter). This will be the high selling camera for Nikon.

- A 45 megapixel camera (or around that amount... but I suspect Nikon will use the same 45 megapixel sensor of D850 with AF system over it to save costs, they already paid for the design and manufacturing fine tunning of it, now it is just ordering a new batch of them), focusing on image quality and resolution, to attract the possible buyers of a Sony a7r III or D850.

I'm not expecting a Nikon megacamera that beats everything... that it is just wishful thinking. Those two will cover nicely the market of people thinking to switch to a Nikon mirrorless system.

That's pretty much my thinking.

I'd also expect the 24MP camera to have better AF than the 45MP camera, since the sensor design is likely to be new and designed from the ground up with on-sensor PDAF in mind (possibly even the same basic sensor as the A7III, with modified filters/microlenses to Nikon's specifications), while the 45MP version will probably use a modified version of the D850 sensor, with more limited capacity to support PDAF (unlike the A7r2/A7r3 sensor, it was not designed specifically for mirrorless use, since the D850 didn't need it). The 45MP version will cost more than the 24MP version, regardless of AF performance, but will probably undercut the A7r3, since it won't be able to compete on lens selection and is unlikely to compete on AF. It will be like Sony's initial launch of the A7 and A7r in 2013, with a cheaper, lower-resolution, higher-performance body, and a high-resolution version that doesn't perform nearly as well but is optimised for studio/landscape use.

Sony will probably do nothing. They released end of last year the a7r III and this year the A7 III. Those cameras will still be on the market for another year before being replaced. By the end of the year we will probably see the A7s III, more focused on video and low sensitivity (the A7 III has a sensor that has a high-ISO noise similar to the A7s II right now..., Sony has to best themselves there). Of course, this camera will need to best the Panasonic GH5/GH5s in terms of video features... that it is their competitor in the SLR video market. Maybe we see the infamous a6700...

They're not going to replace the A9 or A7r3. But they may release an A9r in the first half of next year, combining technologies from the A9 and A7r3 and trumping the Nikon/Canon releases with a 60+MP, A9-level AF, 10fps flagship model. If they do, that sensor will probably find its way into the A7r4 a year or so later - Sony's never been afraid to cannibalise its own lineup.

They'll also release an A7s3, but that's already expected.

Next year we maybe see an A9 II... or maybe beginning 2020. Sony will want to have more people in Tokyo Olympics in 2020. And of course, continuing making new lenses.

I would expect 2020. They would want to launch it in an Olympics year, so as to have a new product to sell rather than a one-year-old product, and there isn't enough time for two generations of the A9 between now and July 2020. Also, they'll want 8k video for it (being the Tokyo Olympics) - a 2020 launch allows more time for camera bandwidth, card storage and batteries to catch up (sensors are already there), as well as users' computers to be upgraded to the point where they can actually process 8k video instead of sparking a torrent of complaints along the lines of 'the files are too big for my computer/my computer's too slow', as would likely happen if they released an 8k video camera now.

In the meantime, they'll likely bring out a few more long lenses (500/4, 600/4 and possibly an equivalent to the 180-400/4 or 200-400/4) to allow the new camera to be used to its full capacity when it comes out.

I highly doubt Sony Imaging Products and Solutions has inside information from the Sony Semiconductors company.

At the board level, they would.

Even if the company itself (outside of the board) does not, they don't need it.

Sony cameras get their sensors from Sony Semiconductors. Sony Semiconductors is constantly developing and testing new prototypes, keeping a 'catalogue' of prototypes that Sony camera (and others) can select from when designing a camera. And these prototypes will incorporate technologies from every source, whether developed by Sony or taken from a Nikon sensor developed in conjunction in Sony - Sony camera doesn't need to know where each individual technology came from, only that Sony Semiconductor is offering them a prototype that uses it.

[BernardLanguillier]

The main logical argument I read here is “Nikon machting Sony in resolution and AF in their first try would make me feel insecure as a Sony customer”.

I have already written 20 comments ago that I thought that matching the D5 would be challenging, but rather credible leaks mention that they may.

For what it’s worth, the Nikon 1 series AF is 7-8 years old tech and was already in DSLR class from a speed standpoint. It seems reasonable to expect Nikon to do better 7 years later in a body costing 7 times more, right?

It seems pretty obvious that the AF is a key area and that Nikon will deliver a very solid performer since they have both had the technology for years and a clear history of delivering the best DSLR AF performance.

Regarding your theory that the high res sensor may have worst AF, we will see. In the DSLR world Nikon has always equipped their high res body with the same AF sensor tech as their sport body. The main reason why the D800/D4 or D850/D5 don’t perform exactly the same is related to different mirror boxes. There is a lot less valid justification in the mirrorless world.

The only one I can think of is the speed of data sampling as it id read off the sensor. But this would be an indirect consequence of resolution, not a direct one. And it would assume that the AF info is read with the same mechanism as the imaging information, which may or may not be the cass. It seems to be a technological implementation story rather than a fundamental limitation. Your guess is as good as mine in terms of Nikon relying on off the shelf sensor AF Sony technology vs developping their own.

In any case, I just don’t buy your explanation about the high res body being based on an existing D850 sensor. Or, more accurately, on a sensor that wasn’t designed with mirrorless in mind.

But we will know in 2 weeks.

Cheers,
Bernard

[D Fuller]

...
But there is no way they will match both the A9 and A7r3 in the same camera this year. It's one or the other.
...

Basing it on the D5 sensor wouldn't help either - not only does it lack the resolution to compete with the A7r3, but it also lacks on-sensor AF, the D5's AF performance being derived from off-sensor components rather than the sensor itself. 

Well, this is all true, but the A9 is no match for the A7R3 in resolution either. It seems obvious from all the real-world cameras that there is a trade-off between autofocus and frame rate on one hand and resolution and DR on the other. The reason for the frame rate trade off would be all about data rates—you just have to move a lot more data at 45 Mpix than at 24. It’ll get worse if DR reaches the point where 16 bits are required to represent it.

Focus speed is more of a mystery to me, but all of the best autofocus cameras are twenty-something megapixels. Can that be coincidence? Even Leica, who doesn’t seem to care at all about how much they have to charge for a camera, made the SL 24 Mpix, so I suspect it’s not down to money.

[shadowblade]

Well, this is all true, but the A9 is no match for the A7R3 in resolution either. It seems obvious from all the real-world cameras that there is a trade-off between autofocus and frame rate on one hand and resolution and DR on the other. The reason for the frame rate trade off would be all about data rates—you just have to move a lot more data at 45 Mpix than at 24. It’ll get worse if DR reaches the point where 16 bits are required to represent it.

I never said that the A9 was a match - if it were, I'd be shooting an A9 rather than an A7r3. I was merely responding to the assertion that Nikon's new camera will compete with both the A7r3 and A9. That would require a completely unprecedented leap in capability - more likely as an evolution of the A7r3 (as the A7r4 or A9r) than from a company which has never made a full-frame mirrorless camera before.

Focus speed is more of a mystery to me, but all of the best autofocus cameras are twenty-something megapixels. Can that be coincidence? Even Leica, who doesn’t seem to care at all about how much they have to charge for a camera, made the SL 24 Mpix, so I suspect it’s not down to money.

It's down to bandwidth and frame rate.

In SLRs, there's no connection between the sensor and the AF system. You can put the same sensor in two different cameras with two completely different AF systems, or vice versa. The 1Ds3 had the same AF system as the 1D3, while the D3x had the same AF system as the D3. You could easily put a D850 sensor into a D5 and get the same AF performance.

But people who need the fastest possible AF often need frame rate more than resolution. And that applies just as much to SLRs as mirrorless cameras. The 1Ds3 could only manage 5fps. The D3x only managed around 2-3fps. Even with top-tier AF systems, if you can only manage slow frame rates, you don't have much of an action camera. And, for a non-action camera, the AF system is largely wasted. It's only in the latest generation of bodies - A7r3 and D850 - that resolution and speed been combined to achieve both high resolution and acceptable frame rates for fast action.

[shadowblade]

The main logical argument I read here is “Nikon machting Sony in resolution and AF in their first try would make me feel insecure as a Sony customer”.

I change systems every few years anyway, due to fairly high attrition from broken gear (I have a 1Ds3 with a bullet lodged in it, and I'm pretty sure a crocodile has my 5D2 with attached lens), thefts and obsolescence. But the next one will almost certainly be either Canon or Sony - whoever has the most promising lineup when I next have a large chunk of gear destroyed - not the comparative minnow which now has to move into a completely new field of camera technology, without the benefit of being an incumbent with large pre-existing market share. Sony made their move five years ago; Canon has far greater resources.

I have already written 20 comments ago that I thought that matching the D5 would be challenging, but rather credible leaks mention that they may.

Where are these leaks, and why are they credible? The only thing I've seen on Nikonrumors reads like some fanboy's wishlist.

For what it’s worth, the Nikon 1 series AF is 7-8 years old tech and was already in DSLR class from a speed standpoint. It seems reasonable to expect Nikon to do better 7 years later in a body costing 7 times more, right?

Only when moving small pieces of glass, tracking slow-moving, simple subjects, with lenses that don't even need to focus all that accurately, since DOF is so large. Quite different to focusing a 200/2.8 lens on a running subject and achieving accurate focus on the nearest eye.

Olympus E-PEEN series M43 cameras can do the same thing. You'd never put them in the same class.

[QuoteIt seems pretty obvious that the AF is a key area and that Nikon will deliver a very solid performer since they have both had the technology for years and a clear history of delivering the best DSLR AF performance.[/quote]

Which does not translate at all to being able to build a class-leading sensor-based hybrid PDAF/CDAF system. They've never even made an on-sensor PDAF system.

In any case, I just don’t buy your explanation about the high res body being based on an existing D850 sensor.

But we will know in 2 weeks.

Cheers,
Bernard

Then what are they going to base it on? And how many different sensors is Nikon going to support at the same time? How many can they afford to develop at the same time? The D850 only just came out.

The A7r3 sensor was amortised over two generations and three cameras (A7r2, A99II, A7r3). The previous 36MP sensor went into a lot of different cameras from multiple manufacturers, with only small differencea between them. And Sony is a huge company with the world's leading market share in sensors. The D850 sensor has gone into exactly one camera, and Nikon is a comparative minnow. If they developed that many sensors, they'd never get their money back.

[BernardLanguillier]

How can you know before product announcement that you will not be using Nikon equipment moving forward?

How has Canon’s greater resources helped them deliver best in class DSLRs to their customers these past 8 years?

When do you think Nikon froze their mirrorless design and how could that have related to the D850 development?

When you write "They've never even made an on-sensor PDAF system.", are you aware that the Nikon 1 series was the first camera ever released with an on-sensor PDAF system? Besides, as was just pointed out to me by Jack, they have owned patents on this for 6 years... https://nikonrumors.com/2017/08/03/new-nikon-multi-pixel-pdaf-sensor-patent.aspx/
...

Cheers,
Bernard

[BernardLanguillier]

https://nikonrumors.com/2018/08/09/third-nikon-mirrorless-camera-teaser-released.aspx/#more-124463

Nikon is saying "we have invested all our knowledge in this camera". Looks like a high end release.

Cheers,
Bernard

[Guillermo Luijk]

Nikon is saying "we have invested all our knowledge in this camera". Looks like a high end release.

To me that looks like "Everything will be fine" from your mom right when you get into surgery, words expected to be heard but zero real information. No camera brand would admit they put partial knowledge on a new development

[BernardLanguillier]

To me that looks like "Everything will be fine" from your mom right when you get into surgery, words expected to be heard but zero real information. No camera brand would admit they put partial knowledge on a new development

Well... never under-estimate Nikon's ability to mess up marketing. 

Don't forget that they released the D3x and didn't say a word about its dynamic range... https://imaging.nikon.com/lineup/dslr/d3x/pdf/d3x_24p.pdf

They are just telling us that it is one of their most strategic camera ever. And coming from them, it tells me something because... as far as I recall, they have always made a big splash when they wanted to/had to. The D1 was the first native DSLR at a moment when Canon was leading in the film world thanks to USM, the D3 broke new grounds in high ISO and AF when Canon was leading in FF bodies, the D3x was the first high DR camera, the D800 broke new grounds in resolution,... there is little historical reason to think that they would mess up what themselves describe as one of their most strategic camera release ever.

All of their recent DSLR lenses have been outstanding in terms of look and technical qualities too. Their new mount should have provided them with all the design freedom they needed as well.

Cheers,
Bernard

[shadowblade]

How can you know before product announcement that you will not be using Nikon equipment moving forward?

Probability. Mirrorless is a new field - all camera companies needed to produce new mounts, new lenses, new AF systems and new ways of doing things. Existing SLR market share has limited value. Canon and Sony are huge companies, with major sources of income outside photography to keep them running during any bumpy transition period. Nikon is almost entirely reliant on still cameras. Sony got a five-year head start. Canon is a much bigger company than Nikon, and has a mature and effective dual pixel AF system (with the potential to become the best AF system out there, even better than Sony's current system, although Sony has recently patented a similar technology). So, with Sony starting out ahead, and Canon and Nikon starting from around the same point, but Canon running with far greater horsepower, odds are that Canon will end up with the more capable lineup and greater third-party support. Nikon will have to pick its battles carefully, spending resources to fill underserved market segments, rather than simply throwing resources into a broad, omnidirectional push into mirrorless like Canon can.

It's not for nothing that, even ten years ago - before Sony had even made a full-frame SLR - Canon named Sony as its greatest long-term rival in the camera field, not Nikon. That prediction is just starting to bear out now.

Odds are that, in five years' time, it will be one of the big two - Canon or Sony - at the top, and not the much smaller Nikon. It's simply down to size.

How has Canon’s greater resources helped them deliver best in class DSLRs to their customers these past 8 years?

They haven't. It's also obvious that their efforts weren't oriented in that direction. Canon have been coasting in the SLR world since the 5D3, if not the 5D2, relying on incumbency, existing market share and brand name to keep the cash flowing while they focused their efforts elsewhere.

What has Canon been working on in the past eight years? Video. Dual pixel AF and other video AF technologies. EVFs. Data processing for video and live view. Video-oriented lens technology. Fast sensor readouts, to support video frame rates.

And, guess what? All these technologies are equally applicable to mirrorless cameras, and will help Canon move forward in that era.

Meanwhile, Nikon has been busy breeding a better horse and more efficient carriage, while Sony's car has gone from walking pace, to running, to now matching the horse, with much more potential for future development, and Canon's been developing their alternate car engine and associated technologies, in readiness to move into that market when the car finally outpaces the horse. Much of Nikon's horse-breeding and carriage development, eking every last bit of performance out of it before it becomes obsolete, won't be applicable when they eventually move to cars, while most of Canon's work will - while Sony decided to abandon the horse at the first whiff of petrol, moving into that emerging market and building up market share and technology before anyone else realised the horse was approaching its use-by date.

It is telling that Canon's new 70-200/2.8 is a bare-minimum rehash of their existing lens, and that Canon didn't even bother teying to disguise that fact. It's just a way to keep SLR users happy and, perhaps, to simplify their manufacturing processes. Their real work is going into the mirrorless 70-200/2.8, with mechanics and electronics optimised for mirrorless focusing. They know they need it for a successful full-frame mirrorless launch, and will have it ready.

When do you think Nikon froze their mirrorless design and how could that have related to the D850 development?

They haven't frozen it. They just haven't proceeded as fast as the other two major players. Sony's bern making full-frame mirrorless cameras for five years. Canon's been using mirrorless-related technologies in their cameras and (especially) video cameras for just as long, with each iteration improving on the last - to make a full-frame mirrorless camera essentially involves repackaging the same technologies into a different body. In fact, the 6D2 can essentially be run as a full-frame mirrorless camera, retaining effective AF in live view mode, quite unlike the Nikon bodies (and Canon non-dual pixel bodies). Meanwhile, Nikon had not demonstrated most of the required technologies at all, and it took them until quite recently to even demonstrate a half-decent live view display.

When you write "They've never even made an on-sensor PDAF system.", are you aware that the Nikon 1 series was the first camera ever released with an on-sensor PDAF system? Besides, as was just pointed out to me by Jack, they have owned patents on this for 6 years... https://nikonrumors.com/2017/08/03/new-nikon-multi-pixel-pdaf-sensor-patent.aspx/

I meant full-frame. Sony's been doing it since the A7. Canon did it with the 6D2. Nikon has yet to do it.

True pixel-based PDAF systems, like Canon's dual pixel, are likely to be the future. Sony recently got a patent for something similar, but this is one area where Canon has the lead over Sony. Make it quad pixel and every pixel becomes a cross-type sensor. Sony's separate layer approach (using certain rows of AF pixels) may be better in dark conditions, but this could be surmounted by binning pixels together for the purpose of AF, effectively making bigger and more sensitive AF pixels.

[shadowblade]

https://nikonrumors.com/2018/08/09/third-nikon-mirrorless-camera-teaser-released.aspx/#more-124463

Nikon is saying "we have invested all our knowledge in this camera". Looks like a high end release.

Cheers,
Bernard

Looks like high-end marketing speak. Every company says something along these lines, for every release. Canon said it every time they rehashed their tired old 18MP APS-C sensor in a new body.

[BernardLanguillier]

Great then, you have nothing to fear, your Sony will remain on top.

I’ll remain fact based when selecting my future mirrorless system. The better photography tool will win, regardless of the brand.

I am the first to applaud the work done by Canon on video AF. But, although their technology is mirrorless based onviously, I don’t think that their algos are so relevant for still. The focusing strategies are pretty different.

But anyway, I don’t really care.

Let the better company win.

Cheers,
Bernard

[davidgp]

Great then, you have nothing to fear, your Sony will remain on top.

I’ll remain fact based when selecting my future mirrorless system. The better photography tool will win, regardless of the brand.

I am the first to applaud the work done by Canon on video AF. But, although their technology is mirrorless based onviously, I don’t think that their algos are so relevant for still. The focusing strategies are pretty different.

But anyway, I don’t really care.

Let the better company win.

Cheers,
Bernard

Ok, I read the new comments very quickly so probably missing something... if that it is the case just let me know.

For me, being a Sony user (right now, I was a Canon user several years ago...), I wouldn't mind Nikon releasing an overall better mirrorless system that Sony, that will really force Sony (I'm not expecting Canon to do too much...) to improve... if they can. If Nikon releases a better system would I change to it? No... economically for me it is not possible. So I will buy a Sony 100-400 in the future (closing the lens that I typically use) and maybe in 10 years I see if it compensates to me to change to other system. But looking forward to what Nikon has to offer and how the technology evolves... to satisfy the nerd inside of me

[davidgp]

It's down to bandwidth and frame rate.

In SLRs, there's no connection between the sensor and the AF system. You can put the same sensor in two different cameras with two completely different AF systems, or vice versa. The 1Ds3 had the same AF system as the 1D3, while the D3x had the same AF system as the D3. You could easily put a D850 sensor into a D5 and get the same AF performance.

But people who need the fastest possible AF often need frame rate more than resolution. And that applies just as much to SLRs as mirrorless cameras. The 1Ds3 could only manage 5fps. The D3x only managed around 2-3fps. Even with top-tier AF systems, if you can only manage slow frame rates, you don't have much of an action camera. And, for a non-action camera, the AF system is largely wasted. It's only in the latest generation of bodies - A7r3 and D850 - that resolution and speed been combined to achieve both high resolution and acceptable frame rates for fast action.

In mirrorless systems there is a connection between the resolution and the AF system. The A9 is able to read the image from the sensor in a rate of 60 images per second, that information is used by the LSI chip in the sensor of the camera to help the tracking algorithms. I'm assuming that this sensor being 24 megapixels "only", they are able to reach that speed... in something like 42 megapixeles they are probably not going to be able to do it (right now). So, less megapixeles, you are able to read faster the image from the sensor and use the whole image to improve the tracking algorithms for things like Eye AF.

[BernardLanguillier]

This is apparently indeed the case with Sony’s implementation, but nothing would prevent a different implementation where the AF pixels are read at a higher speed than the imaging ones (a multiple obviously).

In fact I am wondering whether the Sony implementation is really the one you are describing for the PDAD based focusing phase. It may be the case if a final contrast based AF is needed. But this also is technological in nature.

In other words, I believe that a lot can still be invented here.

Cheers,
Bernard

[chez]

How has Canon’s greater resources helped them deliver best in class DSLRs to their customers these past 8 years?




...

Cheers,
Bernard

Canon took a different approach these last few years by making bucket loads of money, not having to restructure, not having to take huge write downs, not having to close down plants.