Sony forms new AI organization to develop new tech for cameras and more

Well, it seems those phone systems are intended to frustrate and infuriate callers. And in that respect, I'd say AI is phenomenally successful! ;)

Dec 1, 2019permalink

Akpinxit

voice activated image search in Goolge project ?

Nov 27, 2019permalink

George1958

I guess if Sony have not outlined what they are doing, it’s not really an announcement?

Nov 27, 2019permalink

AbrasiveReducer

If it's from Sony and things are slow, it's an announcement.

Dec 1, 2019permalink

George1958

sure Sony is doing........ something?

Dec 1, 2019permalink

DavidArmenPhoto

This is likely the infancy of mobile computational photography slowly coming to ILCs.

Nov 27, 2019permalink

mezastel

Funny because they killed app hosting on the A7 :(

Nov 30, 2019permalink

wasTF

Skynet

Nov 27, 2019permalink

milkod2001

I do like this Ai and computation affords coming into photography but would prefer more work to be done on sensors and hardware side of things. ISO 3000 and up on A7R4 is very mushy. Sony do something with it first and then the rest please.

Nov 27, 2019permalink

milkod - AI/computational photography might actually be the answer to image quality issues. If AI can learn how to differentiate between noise and the subject, it can replace the noise with the colour of an adjoining pixel, thereby eliminating the noise.

Similarly AI will be able to differentiate between the subject and the background, and produce sharper edges without the artefacts (haloes etc).

Also, by using depth mapping, AI and deep learning should be able to "predict" well enough to anticipate how a frontal view of a face would look from the side, and this could be used to turn normal 2D images into pseudo-3D images that have a far more lifelike appearance.

Ultimately this could be all carried out in-camera.

Incidentally, you say that "ISO 3000 and up on a7Riv is very mushy", but mushy in relation to what? Are you claiming it is worse than on other Sony cameras, or worse than on some other specfic brand/model? And on what evidence?

Nov 27, 2019*permalink

milkod2001

I got what you mean but this automation is never gonna be as good as real photo. I sure will help to some degree.

Regarding ISO 3000 and up on a7Riv. It is indeed mushy. Check on many samples on internet available. I think Sony has rushed this sensor and did not optimise properly. They pushed for too many MP sacrificing High ISO.

Nov 27, 2019permalink

milkod - Yes, the megapixel race has started all over again. Rumours are that Canon may launch an 83MP camera soon, but it's just overkill for most people. For me, the ideal sensor would be about 40MP, and designed to produce high DR and minimal noise in the 400-3200 ISO range where I normally shoot.

Regarding AI/deep learning computational stuff, I can see a lot of potential benefits, but the downside is that it could result in lazy, sloppy photography, with little mental input from the photographer, so while the results might be better technically, they might suffer artistically.

Nov 27, 2019permalink

phouphou

i hope this also means, they will intergrate AI like in google Pixel phones.
Considering what those tiny sensors of phones can do with ai should make larger sensors extremely powerful.

Nov 27, 2019permalink

Anders_K

A selfie camera with an auto-beautifier would become a smasher among kids.
Big money there. Then place little ads into the photos and draw royalties, again big money there. Don't forget to send data to Google and other Big Brothers...

Nov 27, 2019permalink

urix Lukin

Why it would be a smasher?
Averyone and their mom have smartphone already.

Nov 27, 2019permalink

Franz Weber

Rumor has it that the upcoming model will have three groundbreaking features:

- It will automatically post your best photos on dpreview,
- react to troll post in the Sony forums,
- and start equivalency threads in the m34 forum.

Nov 27, 2019*permalink

It would be fun if AI created a review and then AI bots were trolling and responding to comment section all by themselves. Its like Alpha Zero vs Stockfish in photos.

Nov 27, 2019permalink

iAPX

@zxaar Chess Player and previously Chess Engine Developper here since the 80's, and your comment is lmao :)

Nov 27, 2019permalink

Franz Weber

They need a reasearch team to develop a inbuilt comfortable grip for their Rx100 line

Nov 27, 2019permalink

They very much need an ergonomics development unit, and a menu development unit.

Nov 27, 2019permalink

caver3d

Now watch Sony and Panasonic continue to duke it out with AI. The innovation will come from these two electronics giants. We're already seeing them build in such technology. The future A7/A9 and GH/S1 cameras will be something to behold. These two heavyweights are circling each other and waiting to land that next uppercut. Will it go the distance?

Nov 27, 2019*permalink

Duncan M

Last Sony release was a big bummer in specifications.
You already seem to have forgotten the A9II DIDN'T bring much and could be seen as a big disappointment.

If that's the future of Sony imaging I would not be too excited about their future cameras to come.

Nov 27, 2019permalink

badi

"If that's the future of Sony imaging I would not be too excited about their future cameras to come."

Give it a rest already, they (Sony) pushed really hard on tech for some long years. Finally the current state of the ML is somewhat mature. Dream AF performance and implementation, better that optical EVFs, dream camera performance and other specs, A9 in the position where many pros would prefer them other the big classics. And all that while constantly pushing the sensor tech as well.

They need to relax a bit and put a harder work on details - ergonomics, menus, touch interface, etc. for a few generations. At the current state, the only thing other then minor performance increments would be huge improvement on sensor read out times (already impressive but with some minor issues on the A9) or, eventually, a full electronic global shutter.

Nov 27, 2019permalink

RedFox88

Artificial intelligence doesn’t exist. It’s all computer programming, not software that can learn or be told what to do!

Nov 27, 2019permalink

Revenant

Machine learning and neural networks are part of the research field called AI. "Artificial intelligence" doesn't necessarily mean a self-aware intelligence with a conciousness and a will of its own.

Nov 27, 2019permalink

TristanW

current real-time-track is really nice, wish SONY to bring hardware-level-HDR, hand-hold long-exposure ASAP. which is so mature in latest mobile phone

Nov 27, 2019permalink

Thoughts R Us

All major electronics companies need to have an AI initiative and so this is should be no surprise.

Nov 27, 2019permalink

thx1138

Sony should worry about UI first, not AI!

Nov 27, 2019permalink

Thoughts R Us

Ain't that the truth!

Nov 27, 2019permalink

the whole UX actually not only UI

Nov 27, 2019*permalink

Well the truth is majority of Japanese companies neglecting UX as a whole

Nov 27, 2019permalink

sony needs to ignore few people and they seem to be already doing.

Like the sony ui. No issues here. They should not change.

Nov 27, 2019permalink

"What don't Sony implement a proper touchscreen first? "

You can always buy canon and play with touch screen, while sony won't need touch screen because AI would do all,

Nov 27, 2019permalink

AbrasiveReducer

Well said. They already have plenty of artificial intelligence.

Dec 1, 2019permalink

nzmacro

So basically to make point and shoot cameras. Doesn't appeal to be honest.

Nov 27, 2019permalink

Thoughts R Us

Most of this most likely will be directed to stuff other than cameras, from Gaming to image sensors for smartphones and cars and surveillance, etc.

Nov 27, 2019permalink

You know if a camera has a really good AF doesn't make it a p&S camera. With one more variable automated you can focus on the content (exposure, composition, color, etc) rather than fiddling/worry with your focus the whole time

Nov 27, 2019*permalink

@bakanecko

I think he means automation in other aspects. On the one hand for newer generation point and shoot types be better , for us (me and it seems him) more manual control side is fun.

I hope we don't lose the manual controls over the cameras. That is only drawback i could think of AI.

Nov 27, 2019permalink

What could i think what we can have benefit from AI, better subject tracking, better auto white balance, better min auto shutter speed, electronic stabilization, reducing rolling shutter, better at eliminating light flicker, maybe a better dynamic range. Which still leave us some manual operation here and there.

Nov 28, 2019*permalink

"What could i think what we can have benefit from AI, "

Recently found out that one of my close friend from college is now Director Machine learning at Adobe. When i talk to him i am curious to know what they think it would be and see where they are working.

Nov 28, 2019permalink

Slouch Hooligan

We now have AI that can sort and rate photos. When we have cameras that are autonomous Humans can be removed from the loop altogether.

Nov 27, 2019permalink

3pgrey

AI could be used to recognise a scene for intelligent stacking

Nov 27, 2019permalink

pentaust

The best use of AI robots could be to participate in dpreview forums discussions to save time to real human beings.

Nov 27, 2019permalink

rubank

I have absolutely no need for artificial intelligence,
my own is more than good enough.

Nov 26, 2019permalink

Slouch Hooligan

Maybe Sony will direct Skynet to sort out its menus, ergonomics, and haptics🙂

Nov 26, 2019permalink

onlyfreeman

The Sony a7R VII with Skynet AI built in, will come to the conclusion that the menus are fine, but humans are the problem. We all know how that story ends.

Nov 26, 2019permalink

Lol, 7RM7 won't even listen to you. It'll tell YOU what to point at! Then it'll take the picture how it wants to.

At least it'll solve the Sony ergo/haptics/menu problems. There won't be any controls or menus. Hell, the 7RM7 will probably have gyros that just point itself, and then shoot on its own. ;)

Nov 26, 2019permalink

Maybe when "Skynet" Sony cameras become self-aware, they'll be embarrassed by their own ergonomics!

Maybe if they don't like you overriding their choices, you'll be terminated...

Nov 27, 2019*permalink

Heheheheh, so you think you'll still have the option to override... Hold my beer... I'll be baaack. ;)

Nov 27, 2019permalink

solarider

How about a UI Simplification Organization, too? :-P

Nov 26, 2019permalink

This is, in many ways, a better move than what most Western companies do, where they distribute AI R&D across divisions/labs out of fear of regulations, often resulting to overlap.

I actually met Prof Hiroaki Kitano, who described to me and others over dinner how Sony has been investing to AI R&D way before the current AI hype circle. He talked a lot about his passion on AI for healthcare and robotics. I never thought that Sony will take the next step to create a separate division to consolidate efforts within their vast organisation.

I am particularly happy about the inclusion of AI ethics even in the press announce. Not only as this is my current research domain (and what I discussed about in a panel with Kitano when I met him), but also as it signifies their importance to Sony AI.

Overall, this is an amazing announcement and not just for photographers! It needs to be said, Kitano is a visionary and for a long time was chair of IJCAI. I am very happy that he is leading this.

Nov 26, 2019*permalink

Yes, passion for ai in health care and robotics...
Remember that when things will go out of control.
So short sighted. So unbelievably naive.

Nov 26, 2019permalink

@Marek R: I have no clue what you are on about. However, you can have a look at his work in Nature about drugs discovery.

Nov 26, 2019permalink

You could choose to not mention fact you was around 'world class professor' but decided to mention him and yourself in one line.
You feel more important than you really are, i say fine, but both, you and him, are very very naive having any hope in ai and its use in human problem solving on large scale.
On the other hand, when i think about it, now, ai my solve our problem for good...
I understood why book from thousands years ago was telling is that htruth won't find it's way to people of 'science' but instead will appear for little ones.

Nov 26, 2019permalink

Being in academia frequently puts you in touch with "world-class professors." If the non-intentional lack of humbleness offended you, however, I am sorry.

Having said that, actually intelligent systems already have been helping us at identifying large-scale problems. It is not the magic solution to everything, but as a good ML system can do pattern matching over incomprehensible amount of data for a human being, it opens new doors at research, risk prevention, prediction, etc. A good example is the work done at Stanford at using satellite data to detect drugs, droughts, and help farmers in 3rx world countries plan over the year. In case of his work, ML is used to identify genes and simulate drugs, helping to accelerate the process.

Nov 27, 2019permalink

Sorry, i did not meant to sound offended by unintentional lack of humbleness rather used it to contrast hope you mention with it.
I was not taking about inteligent programs and it's use in current problem solving applications.
I talk about 'final' application of ai which is desire of scientific and not only community that has its hope in technology as savior tool.
It will be used against us and proves are overwhelming.

Nov 27, 2019permalink

User7440441409

@marek, I don't understand what you're getting at. Are you saying that AI won't be used to try to solve large scale problems? If so, then you are the one being naive. It's being used to analyze data from the subatomic level to planetary level as well as doing mundane tasks such detecting cats in videos.

However, if you're saying that @athonline is being naive to think that AI won't be used to do a host of negative things, while I'd completely agree that there are many nefarious uses for AI, I don't think @athonline was implying that AI would only be used for 'good' or that even research done by himself/Kitano-san might not end up being used for bad purposes regardless of initial positive intent (e.g. cat seeking missiles).

Nov 27, 2019permalink

I think Marek's fear is that a massive company like Sony getting into AI big time might have unforeseen consequences.

To keep it simple, let's just look at photographic applications:

Benefits might include much more accurate metering, subject recognition, AF acquisition and tracking, which frees the photographer to concentrate on composition and timing. AI will soon be able to entirely eliminate image noise. Before long cameras will be able to recognise, identify and label images in real time. And you'll be able to disable AI if you choose to.

But there are big risks too. AI has already been demonstrated to be capable of creating convincing video of non-existent events. Faces can be superimposed, facial movements can be manipulated, and speech can be faked. Terrorist organisations and enemy states will be able to produce apparently genuine video of entirely non-existent actions.

Nov 27, 2019permalink

User7...@
There is prospect of ai applications sold to us, the public, and real goal for ai, known to some and deducted by some others.
Of course ai will help solve many smaller but vital issues and slowly is doing it now.
And public will see it is goooood.
But, later, true real goal of ai creation will become apparent...and it will be too late to stop it.
Futurists like kurzweil or ai pros like goetzel are not even hiding their real 'hope' for ai and its role.
They hope to create a sort of super tool that will help sme become immortal on one hand and help manage humanity as problem on the other.
Of you read about ai through scientific publications only it may not be so obvious right away but if you will take step back and looknfrom a right angle and have richer knowledge about humanity's history and ideologies you should see ot clear as night that is ending soon.

Nov 27, 2019permalink

3pgrey

@athon
AI ethics sounds interesting. AI is definitely a topic that needs to be explored and talked about before it can really be embraced as a society in a fundamental nature. Its good to see this announcement from Sony, but I don't envy the hard work it must be

Nov 27, 2019permalink

Entoman@
Components of ai future success are being spread and divided into pieces - sections matrix.
When all will be ready, you can build humpty dumpty again, and missing parts will be solved by itself.
In order to give ai chance to acquire singularity which translate to awarnes of itself more or less, you have to create simple - complex program enough and give it platform of enormous data so it can digest it in order to 'open its eyes' to its existence.
That is why 'they' were gathering all this information from all over the world.
When you accomplish this goal, you have conscious program that can run whole world at every level of activity that depends on network(s).
That is why 'things' were and are transformed into 'smart' status.
Tbc...

Nov 27, 2019permalink

...
You need complex grid of powerful energetic grid that can not only connect smart things wirelessly but even co-power the crucial ones.
If you have drones and robots to use, you can do what ever you like with the life and destiny of sub species that are connected to the network of smart things as last component of wonderfull world.
You can do much more, something so sinister that it is justified to have fear for this kind of 'punishement' let's call it.
And people will buy it, they will love it, until it will be too late.
Proud humans thought they can reach the sky and become like gods...
Ahh , can't help one another yet but are willing to create new species and whole worlds and play gods...
That is the reason punishement will be done with our own hands.
And humanity was and will be shortly forewarned one last time.
Its not far, not long.

Nov 27, 2019permalink

@User7440441409: Exactly, I am well aware that all technologies could (and most are) maliciously misused. In fact, one of the governance requirements I promote in my talks is the consideration and minimisation "dual use"; i.e. when you create something for X, but someone maliciously uses it for Y purpose.

@Marek: I am lost; if you worry about malicious use, then there is not much you can do. Having said that, if Sony AI is indeed putting ethics at the core of its research ethos, you should rejoice. While no AI governance mechanism, whenever that is an internal policy or legislation, will completely remove risks (malicious usage, deception, creation of wrong mental models, etc), they all help minimise or mitigate these risks.

1/2

Nov 27, 2019permalink

@Marek: However, on a second read, what it reads to me is that you are afraid either of "Artificial General Intelligence". This is, thankfully, not a realistic concern.
See, for example, this paper which proofs how not everything is learnable https://www.nature.com/articles/s42256-018-0002-3. Even for the "simple" models ala AlphaGo which populate the media require thousands of year of training (they are in parallel) and the electricity of a small city to train. AGI is (a) impossible from a technical point of view; (b) impractical.

What we should worry about, if I may take a note from Prof Joanna Bryson's talks, is the existing "superintelligence", i.e. the human collective, which is driving other species to extinction by converting all land to houses, farms, or livestock ranches.

2/2

Nov 27, 2019permalink

@3pgrey: I definitely agree. :-) It will require them to produce policy documents and a mind change for engineers.
If you are interested to the field, by all means, please drop me a message and I will gladly point you to papers/books.

Nov 27, 2019permalink

Athonline@
So many time some scientists were claiming that this or that is impossible just to be proved wrong next day.
Luxury ocean liners are impractical same as mega cities that needs sick amount of energy at terrible costs to the planet and humans and both exists.
All it takes is one day when another news will circle the world and new braking discovery will debunke some scientists opinions.
GAI was always main goal of ai promoters and will be here soon.
You will be coughed surprised but that is always the case when someone is standing to close to be able to see something bigger.

Nov 27, 2019permalink

Marek - Ok we all understand your fears, which in the long term are almost certainly justified, but we can't "undiscover" AI, anymore than we can "undiscover" atomic bombs. The cat is already out of the bag and you can't put it back in. The priority now must be to try (and it will be a near-impossible task) to keep AI under our control and not allow it to develop self-awareness.

But as is usual in these forums (and I'm as guilty as anyone), we have strayed completely away from the subject - which is how AI will be implemented in *photography*. How in some ways it will benefit photographers, and how in other ways it will be detrimental.

Nov 27, 2019*permalink

@entoman: True! From what I have seen and heard from colleagues in Computer Vision (CV) and Signal Processing (SP) fields, this ongoing AI hype opened so many new doors and opportunities for their research.
CV researchers are constantly working in denoise, resolution enlargement, further content-aware pixel manipulation (not just deleting, but also exposure adjustments), etc which could influence the quality of our final image. Work in object identification, as we have seen from eye-detect AF and the recent E-M1X and A9 updates can significantly improve AF. Then, there is also the recent trend for image manipulation to construct VR and AR images.
SP research, which often underlies thresholding, can lead to both better analog-to-digital conversion (hence, less noise to start with) or even further advancement of AF algorithms (CDAF works with thresholding and most object identification and tracking algorithms).

In genral, interesting times ahead!

Nov 27, 2019permalink

Entoman@
Maybe its subconscious thing, i don't live in fear, it would be good however, if humanity would start fearing fruits of its labour.
No question about the fact that ai development can not be stoped at present same way you won't convince world to stop aggressive consumptionism so humanity could free itself from being slaves to its own desires.
I did mention that ot will happen anyways.
What is at stake however is to convince as many as its possible to not get themselves connected.
And that is one of the battles of tomorrow.

Nov 27, 2019permalink

Entoman@
Photography?
Its gone.
There is no real reason for its existence anymore.
It will stay here for a little while but its finished really.
In beginning of 2000 i was in love with it too.
Now?
Meaningless.
Advertisement - selling lies, illusion.
Weddings - need for creating false impressions and pose to be someone important regardless of the fact that most couples are investing in material aspects of life without real need to live righteous life.
Filming it with hollywood quality is calculated for an 'effect'.
Instagram, facebook?
Erosion of real meaning and values that ones were so crucial societies coudn't exist without them.
All joke now...and nobody seems to care, nobody seems to notice.
People are busy, black friday os comming, right for their black souls.
😉

Nov 27, 2019permalink

Entoman@
You missed the point with 'not allowing ai to get self conscious'...
This is exactly the big goal of leading developers.
They don't even hide it.
It's all there.
And natural tendency of humans to swimm with the flow, to be ignorant, as it is bliss, is just making it so easy to do.

Nov 27, 2019permalink

badi

@ 'not allowing AI to get self conscious'

Heh... the big trick is that this is such an intriguing box, that it is just too hard to let it there unopened. We are not there yet but one day, and not even that far, self aware AI will be created. And the guys able to do it will do it. I know that I would be too curious to do such a thing if i could.
Now it really depends on what is really possible and what not, and related to this the only certain thing is that even experts opinions are rather speculation based on current tech and current experiments.

Almost 2 years ago, the full neural network of a simple worm was simulated, and apparently is was fully living a virtual life :) https://phys.org/news/2018-02-worm-uploaded-pole.html

Nov 27, 2019permalink

AshleyMC

Excellent!

Unlike the competition, Sony is not shy of non-hardware technology.

Nov 26, 2019permalink

caver3d

Not quite accurate. Do not forget about Panasonic.

Nov 27, 2019permalink

Duncan M

Panasonic can not even bring its AF up to par... solely offering contrast detect AF seems like a poor offering for sports and action photographers.

Nov 27, 2019permalink

PhotoRotterdam

If you want your top of the line camera be turned into a point-and-shoot operation, rejoice.

Nov 27, 2019permalink

And now we know how Skynet will be born. ;)

Nov 26, 2019permalink

Krummj

The end is nigh.

Nov 26, 2019permalink

https://m.dpreview.com/news/7269086157/nikon-invests-in-computer-vision-and-deep-learning-startup-wrnch.amp

Nigh, nay. Here. Simply not instant, that's all...

Nov 26, 2019permalink

mick232

And again, one step ahead of the competition.

Nov 26, 2019permalink

BWfoto

Apple and Google are centuries ahead of Sony

Nov 26, 2019permalink

mick232

Right, but I was implicitly referring to the direct competition in the traditional camera market.

Nov 26, 2019permalink

ozturert

https://www.imaging-resource.com/news/2019/04/05/canon-says-lots-coming-in-2019-lens-strategy-future-of-ai-and-more

Or maybe late...

Nov 26, 2019permalink

EthanP99

Or maybe, still ahead, considering how advanced sony tracking is vs competition. Forming a group later on is not late if the tech is already done.

Nov 26, 2019permalink

Sony is a few years ahead of Canon and Nikon in this area. Clearly. No press releases or Marketing notices changes that.

But if it makes you feel better look for more of that stuff.

Nov 26, 2019permalink

A Owens

Actually Nikon invested in an AI company a few months ago.

Nov 26, 2019permalink

Patlezinc

I don’t think Canon is years behind Sony to be honest. The gap has been quickly filled. Last days there are a lot of videos on YouTube redoing the A7/EosR match and the R wins a lot of fights in photo department. Color science always pointed, among other things.

Nov 27, 2019permalink

"Color science always pointed, among other things."

coated with snake oil??

Nov 27, 2019permalink

A Owens

Color science? DPR rates Canon as the best crop sensor right now. They will no doubt use this sensor as the basis of their next FF sensor. And both Canon and Nikon's recent lens releases have been spectacular. Whilst the Sony system is great and currently leads the way in tracking, other systems are also superb and better in other key areas.

Nov 27, 2019permalink

I think you guys talking about color science don’t know wait AI means.
This article is about artificial intelligence.

Think subject tracking, and subject identification.

Nov 27, 2019permalink

mick232

AI can do much more than subject tracking. It could intelligently sharpen certain parts of the image, it could brighten dark faces, it could correct distortion even without lens profiles, add background blur even with slow lenses, remove chromatic aberrations, remove haze,...

Nov 27, 2019permalink

Sacred

I'm telling you, Sony is our version of Skynet. The Terminator is inevitable reality.

Nov 26, 2019permalink

Its not sony or any specific company but the whole group of them world wide.
Stages and divisions, internationally processed.
Block chains and crypto currencies, face recognition programs and finger print aps, iris recognition and voice control...all information gathered about humanity from ss programs like fb, google, forums, the whole internet with it's access to knowledge about humanity.
After ai will digest all this info, not only it will achieve singularity but it will know humans better than we knoe ourselves.
And than it will end one race to start a new.
Why do you think they call ai and robotic events like 'rise'...and why do you think they were telling this story in books and movies for so long...
It's coming soon.

Nov 26, 2019permalink

https://www.imaging-resource.com/news/2019/04/05/canon-says-lots-coming-in-2019-lens-strategy-future-of-ai-and-more

As usual, Sony leads in imaging technology.
Kudos from a Canon user.

Nov 26, 2019permalink

ozturert

Think again :)

Nov 26, 2019permalink

ozturert - I was simply acknowledging Sony's long-standing commitment to innovation, not attempting to start a brand flame war.

It will be very interesting to see how Canon, Sony and various other brands adopt and implement AI technology.

Nov 26, 2019permalink

Jeff Greenberg

=
entoman: As usual, Sony leads...from a Canon user.
entoman: not attempting to start a brand flame war.

Uses one, compliments another.
Reverse psychology trolling.
How magical...

Nov 26, 2019*permalink

Greenberg - What on Earth is wrong with the owner of one brand acknowledging the work done by another brand?

I criticise when I feel necessary, regardless of whether it's a Canon, Nikon, Sony, Pentax or anything else.

And I give credit where it is due, regardless of brand.

If you see something wrong with that, you have a very warped mind.

Nov 26, 2019*permalink

Ozturert. 2019 is almost over. So what has Canon delivered in terms of AI?

Nov 26, 2019permalink

@Jeff:
Where the hell did that come from???

This is a first. Diplomacy = trolling. Interesting. Tell me you're joking, Jeff.

Nov 26, 2019permalink

Jeff Greenberg

=
> Where the hell did that come from???

It was all about creating a path to using
"How magical" in epigrammatic fashion.
But...

Nov 27, 2019*permalink

ozturert

MikeRan, those investments will find their way. Don't worry. All brands invest in AI now. Olympus' EM1X is the first camera that implemented AI. Next generation cameras will be better.
I meant, Sont isn't the only company and definitely not the first one to invest in AI. Even Nikon purchased and AI company.
But fanatics will be fanatics...

Nov 27, 2019permalink

Oh. Because that article said lots was coming in 2019. Not we’re gonna work on stuff in the background and not release anything.

And yes Olympus has been working hard on AI. But subject tracking as implemented and EyeAF are definitely AI driven technologies. And, well, you know who’s leading there today, even though you would never publicly admit it due to your own biases.

Nov 27, 2019permalink

Huawei? :D

Nov 27, 2019permalink

User1234567890

As long cameras dont make artificial looking photos like samsung phones.
No problem.

Nov 26, 2019permalink

Jeff Greenberg

=
Waiting for BCD function...
(Better Composition Detected)

Nov 26, 2019*permalink

antares103

Its sony, so it will be SBCD (Super Better Composition Detection).

Nov 26, 2019permalink

It's Sony, so it will be PNSD (Point-'n-Shooter Detected). Full-Auto ENGAGE! ;)

Nov 26, 2019permalink

Waiting for SBCD to get so far that you just have to take the camera with you, maybe have it hanging by your side, and it will automatically detect a good composition and take an image.

Photographers will then just be the transport for the cameras. In goes food, out goes CO2, and the camera is happy. Because of course AI will reach that level and have emotions.

Nov 26, 2019permalink

steelhead3

If you have no new products on the horizon...reorganize LOL

Nov 26, 2019permalink

User1234567890

A7 iv on the way.

Nov 26, 2019permalink

Becksvart

A7 ai

Nov 26, 2019permalink

Yes because Sony hasn’t released anything recently.

Nov 26, 2019permalink

First! 😜

Nov 26, 2019permalink

Are you not going to make a comment about Sony's AI?

Nov 26, 2019permalink

Skynet... ;)

Nikon bought WRNCH for deep-learning AI. Canon's got their fingers in AI too. Oly with their train-tracking AF, etc.

Sony is not alone.

Frankly, I don't like it. This creeps me out way too much. Who wouldn't love the convenience of perfect subject-tracking with virtually no effort? But at what cost...

Canon's eye-controlled focus from forever ago is what I'd like to see modernized. Of course, then I'd have to worry about the security issues of my camera retina-scanning me and that data going wherever it will.

AI is the opening of Pandora's Box. And, well, that Box has been opened. No turning back. Pointless to rebel against AI now. But there's too much guilt involved for me to embrace it...

Nov 26, 2019*permalink

Aegon - I agree that all new technology can result in undesirable scenarios, but there are also many benefits to be gained.

Many of these are things that I have no use for (I'm a control freak so I like to make at least *some* of the decisions myself).

But I'll certainly make use of e.g. animal eye-AF, when it becomes more efficient.

The only thing I'm concerned about is that I might not have the option of overriding the AI automation.

For many uses AI will be beneficial (e.g. automating AF and metering to free the photographer to concentrate on composition and timing), but I also strongly suspect that AI will make us all lazier....

Nov 26, 2019*permalink

@entoman:
Yeah, Animal Eye AF is something I'm looking forward to. I'm doing fine without it, but I expect in time it will be better than I am.

Don't worry about AI override for AF. I expect the full palette of "standard" AF-S and AF-C modes will still be the base for Nikon and Canon AF systems, and that the various specialty AF (Eye, Animal Eye, Trains (lol), etc.) modes will need to be selected/activated. Surely they'll always be deactivateable. Afterall, we can still activate full manual focusing. However, Nikon's deletion of AF-C 3D in MILC is a head-scratcher. But not really, as it's pretty clear Nikon was very careful to not cannibalize their DSLRs with the Z6/Z7. We'll see in future iterations.

Hopefully Nikon develops "Hottest-Girl-In-The-Frame Eye AF" first. ;)

I'd still rather have the camera watch MY eye, and focus on exactly what I'm looking at. Because even if a camera's AF can lock on an eye and never let go for a moment, what if I wanna shoot something else...

Nov 26, 2019permalink

Aegon - Yes I'm sure cannibalisation was an issue. They'd have to be pretty daft to make their best ever DSLR obsolete by making the Z7 equally capable. Canon followed the same pattern, with the EOS-R - the poor man's 5D4.

Both the 5D4 and the D850 are looking a bit long in the tooth now, and unlikely to be replaced with other DSLRs, so I expect the next iteration of MILCs to be much better specified, as there will be no cannibalisation issues, and they'll want to match or beat Sony a7Riv.

Nov 26, 2019*permalink

Oh, I think the 5D4 and D850 will have DSLR successors. I'm not giving odds on that one, but I'll take even money bets. ;)

Nov 27, 2019permalink

@Aegon “ Who wouldn't love the convenience of perfect subject-tracking with virtually no effort? But at what cost...”

Are you in photography more for the end-result, or more for the process, tools, etc?

When I first tried out Nikon’s 3D tracking, it honestly felt like cheating. But then I realized that the end-result is what matters most to me. So in the end, the choice was an easy one.

Nov 27, 2019permalink

@starbase:
I feel Nikon's AF-C 3D is the lazy way to go. I can easily outperform it using AF-C Single Point or AF-C Group, and usually with AF-S Single Point.

To answer your question, the answer is "both". I care about the end result, and I care about my actual skill in creating the end result.

Think about it. If the ONLY thing that mattered was the end result, put down your camera and go to a photo art gallery. Just buy the photos you hang. Because there is always someone better than you making better photos than you.

It's like racing a car. I always turn off all assists that can be turned off. Who cares about a lap/drag time that's partly/largely due to computer controls? Not me. It reminds me of my test drive with a Tesla P100D with Ludicrous Speed activated. It takes no brains and no skill at all to launch that car to 60mph/100kph in about 3 seconds. Because of AWD and computers. Fun, but my mom could do it. Let's see mom do that with a 700hp RWD with no traction control...

Nov 27, 2019permalink

@Aegon "I feel Nikon's AF-C 3D is the lazy way to go. I can easily outperform it using AF-C Single Point or AF-C Group, and usually with AF-S Single Point."

Nice. Can you also do that while paying attention to composition, foreground/background, moment, light, etc? In other words, the things that actually make a photo better? If you can, again, nice. But when I can trust the camera to handle focus I find that I can "focus" on the other (and arguably more interesting) stuff better. Which means that at the end of a day, I have more keepers. Which also means I am more motivated to go out and shoot again. Which in turn improves my skills - the ones that matter.

"Think about it. If the ONLY thing that mattered was the end result, put down your camera and go to a photo art gallery. Just buy the photos you hang. Because there is always someone better than you making better photos than you."

Maybe I should have said "I care about my end-result", not "the end-result".

Nov 27, 2019*permalink

@Aegon Besides, photography should be fun. I can determine for myself what that is. What I find fun. How I get the most fun out of it. If you or anyone judges me, that's not my problem.

I am sometimes so surprised by this. By photographers (at least, I hope that people who come here are photographers) talking in absolutes and arguing that other people should do this or that. Hello, it's photography. It's individual to start with. The whole point is to capture what you see. Not what anyone else sees. So you have to decide on how you want to do that. Not anyone else.

I guess this also answers your argument about buying a photo someone else took...

But if I can hold a mirror up: if you are so determined that you want to do everything yourself, why do you even use AF? Why do you (probably) use a light meter, or even - God forbid - auto-exposure?

Nov 27, 2019permalink

I think it depends a lot on what you're shooting, and why you're shooting. Sports photography, or even photographing active kids, means very rapid and erratic movement. With these types of subject you simply don't have time to twiddle with joysticks, or to focus-recompose, or focus manually.

On the other hand, *if* the subject is fairly static, or if its movement is predictable, I prefer to use single-point AF-S and to use the joystick, because where possible I prefer to decide for myself where to place the AF point.

Nov 27, 2019*permalink

@starbase:
Well, I'm not in the mood for a flame war. And if our discussion escalates beyond its current tone, I'll excuse myself from it. That said...

Regarding your "nice" and second "nice", my answer is pretty much "yeah, no problem".

For probably my first 20 years of photography there was no digital photography. With film you either learn to make each shot count, or go bankrupt. For my first 5 or so years there was no AF. These force technique and skill. If you didn't learn that way, I will bet everything I own that you don't shoot the way I do. Nor does anyone say you should or need to. But what you apparently find surprising or likely false, is rudimentary for me. You'll also find that professional sports photographers shoot the way I do. Not with AF-C 3D, and not with Eye AF, or other whole-frame AF tracking. This may change soon. The a9 is quite good at tracking. With some improvement, it'll be pro-caliber. But it isn't yet.

Have fun however you like. Did I say otherwise?

Nov 28, 2019*permalink

Oh, and even for sports and manual focus, it's possible. Of course fast sure AF will provide better hit rate. But when shooting sports manual-focus, you employ techniques learned-on-AF shooters typically never heard of nor thought of.

For instance pre-focus. You'd be surprised how when you know the game and teams/players you can often figure out where the shot's going down. Sometimes you know the money shot will be at the 30 sideline. Point there, focus, wait for it. Things like this. When each frame of your 40-exposure roll costs money, and TIME changing rolls, there is no spray & pray unless you have unlimited budget. You gotta THINK of how you're gonna get that shot.

I suggest everyone try shooting action in manual-focus with 1fps or maybe 3fps drive. Preferably in full Manual Exposure. Then you have to learn which angle and part of the field's EV is and adjust settings as you move the camera. No time to look at the meter. Know you spot and turn the dial as you turn the camera.

Nov 28, 2019permalink

Aegon - All extremely good points.

Nov 28, 2019permalink

Thanks, entoman. Ah, nostalgia... 😉

Nov 28, 2019permalink

@Aegon I didn't mean to start a flame war. If anything, if you really are that good, you know, wow. But it's not for everyone. I'm not a pro photographer. I have a 40 hour day job. Photography is something I sometimes do besides that. And yes, I learn. Maybe in a different way from you. But then nowadays, I think most people are.

I know about pre-focus. I used catch-in-focus myself a number of times. What I don't like about it is that I don't have control over when to shoot. Not saying it's not useful in some circumstances. But as you say yourself, this works "sometimes".

I have been "accused" of using spray and pray before, when I set apart my reasons for switching from Pentax to Nikon. If anything, a more reliable tracking AF allows me to pick the moment in a better way, so that I don't actually have to spray and pray as much. I even tried to explain that...

The buffer on my D7100 is way smaller than that of my K-3 as well, so I can't even really spray and pray as well as I could.

Nov 28, 2019*permalink

@Aegon Also, I use manual exposure a lot. Sometimes with spot metering. Also, it's not as if having an advanced AF makes it impossible to use pre-focus. And sometimes it's the best way. But for me, often, it isn't.

With regard to 3D tracking, I'd imagine most sports/wildlife photographers use something like D-9 or D-21 instead as 3D tracking isn't always the most reliable option.

Nov 28, 2019*permalink

@starbase:
Well, i don't know about "wow". That's just how a lot of us learned. I was fortunate in that within a year of starting photography i owned Nikon's flagship body/drive and a few really good lenses. I also joined my high school's yearbook/newspaper, so therefore got the benefit of the school's supplies/darkroom. Because of that photography i got a free summer before 12th grade of instruction with my State's #2 photojournalist. He was still shooting Nikon's prior flagship body/drive and that served as a constant source of playful insult both ways between us. I read tons about photography, subscribed to the magazines (lol, no internet back then!) before i ever met him. And all that time, often 1-on-1, with a photojournalist of that caliber dishing the dirt was invaluable.

Fast-forward to DPR Comments and, NOT that you're one of them, so many here "learned photography" by playing with their phone cameras then buying a Sony, and now preach the Sony gospel. And I'm the heretic. 😂

Nov 29, 2019permalink

And like i said, whatever works for you works for you. And that's cool by me. My needs are a lot different from most people's. The features that Sony fans can't shut up about mean nothing to me, and what Sony fails to address year after year is what does matter to me, that's all.

But when i hear all this Sony mantra about how the tech is so great yadda yadda, i just laugh and smh. By and large what that tech does is exacerbate phone camera "technique", but with better IQ. Reinforces laziness. When what most Sony shooters, again I'm not saying you, need is some photographic instruction and practice shooting in full manual everything. NOT sophisticated automation...

Nov 29, 2019*permalink

Aegon, you'll get struck down by The Curse of Sony Warriors!

Nov 29, 2019permalink