Cary Knoop
Veteran
They are certainly very inventive, but the question is will they put quality first and try to build a 100-year reputation or are they going for the quick buck?
C100: Canon layoffs
- Thread starter scorsesefan
- Start date
- Status
scorsesefan
Veteran
I think there will always be a market, albeit niche, for full frame, interchangeable lens cameras and cinema cameras. Just as in the pre-digital era people bought Hasselblads, Nikons and Leicas even when Kodak introduced the Polaroid Instant camera.
There are, as usual, countervailing trends. The processing power and the sensor technology is moving forward largely based on the need of the smartphone industry. But the profits also gravitate toward the smartphone industry. Hypothetically, everything that the pros do can be done with a camcorder with a 2/3 ENG, 1" or M43 sized sensor. The ENG camcorders will have the reach, the cinematic brethren will have the wide angle 20-70. They'd also cost about $1,500 for the high end pro version and will sate every gearhead addiction on a low/no budget.
In terms of the global photo/video industry, it's likely to see downsizing, restructuring, mergers and bankruptcies. That Huawei P3 has a 32 MPX sensor with a 50X zoom and 409,000 ISO performance on one of its camera modules. Whereas, prior to now, the battle was of the performance vs. the convenience, the unfolding battle will cross over into the performance territory. An ILC with a 200 MPX sensor and a long zoom will naturally still perform better but the difference of 8K vs. 20K is rapidly becoming irrelevant. A human eye can't see it anyway.
Netflix is borrowing a cool bil because they want to outspend others into an immediate oblivion. Their investment into programming is $15B/Y, almost as much as its annual revenue of ~ $16B.
But Disney, Google and Apple have lots of financial muscles as well. Studios and production houses will have to align themselves with one of them or face an outright extinction.
Except, even the cameras are moving into a "disposable" niche. "Gimme a couple of USB cables, a CFexpress card and whatever ENG camcorder you have on sale. I think I left mine in the car and feel too lazy to walk the length of a parking lot to retrieve it".
Last edited:
puredrifting
Major Contributor
Feels like we are definitely heading into another time of huge change for us, doesn't it?
Creating quality content will always be a valuable sought after skill, regardless of what happens with the tools.
As you know working in LA, many people acquired really high end tools- Reds and Alexas etc. (so called "Trust Fund" producers (TFP)) willing to work for practically nothing, which pushed down the amount people were willing to pay for production work. However that's not a sustainable business even for the TFPs and similar, and thus those that can produce quality content (on any camera) will still be valuable and can keep on truckin' long after the TFPs disappear.
Looking at the threads for this forum over the years, it's been mostly gear discussions. Perhaps it will shift to filmmaking, perhaps the original intent of this forum?
Consider another line for your sig: It's not the gear, it's how you use it that's important.
puredrifting
Major Contributor
Hopefully true. Yes, all of the gear talk does get boring and pointless at some point, they are just tools and really all that matters is what one can accomplish with them. FWIW, a few years ago, I did have some clients tell me on set once, "That looks too good. Can you make it look more like badly shot cell phone footage?" and I did have a couple of clients who though that Go Pro footage looked better than the same shot photographed with a RED, but in the end, all that matters is if clients will pay you to help realize their vision.
F35 was $250,000 back in 2008 and now your XT-3 takes more detailed images for $1,500. The new $7,000 Canon XF-705 seems to have everything one needs for the run&gun/ENG, aside of some file conversion/codec issues. Even its focal length equivalent of 25-380 is sufficient for most uses.
PS. Jared "Fro" Polin reviewed Nikon P1000 and said ~, "It's a mediocre, smartphone quality stills camera but an amazing camcorder due its mega zoom". Sitting in the outfield bleachers for a Phillies game, he was able to do a closeup of the home plate. For $1,000.
puredrifting
Major Contributor
I agree with what JCS says, the future of low end and even into medium end work is with phones. You cannot deny computational power can overcome almost all of the issues behind phones as serious image making machines. Stabilization, DOF, wireless multi channel audio, it can all be overcome with exponential computing power. All of the things that make us use 25lb C300 MKII rigs will be laughable in just a few years.
I was thinking about a "perfect trip" camera of the near future (before we could just store our visual memories in a chip implanted into our brain). My idea would be something of a two sensor-two lens version of RX-100/RX-10 type, where one could use a smaller (2/3 or 1") sensor with a long zoom during the daylight hours and a larger sensor (APSC or full frame) for the low light or indoor shooting. The sensors themselves are getting cheaper and cheaper but a dual cam can share the same brain, spreading the costs around.
And the pro portrait shooters would have have something resembling a classic Rolleiflex with two medium format sensors with a mini-zoom each or even two different primes.
And the pro portrait shooters would have have something resembling a classic Rolleiflex with two medium format sensors with a mini-zoom each or even two different primes.
Maybe we should stop calling them phones? They're now really a micro-portable computer with a modem, a moving IoT node. For most people, the majority of use time isn't the phone. The built in optical sensors (aka cameras; iPhones have an IR TOF camera for 3D depth sensing) will get much better, and decoupled optical sensors could be interesting too (think USB cameras with PC/laptops). While integrated computational sensing can provide arbitrary DOF and lens simulations + any bokeh and pleasing (or unpleasing!) distortion you can imagine, it's still a few years away. Right now external advanced sensors can be used over USB3.x which would turn these mobile computers into much higher quality image capture devices, including using full frame (or any) lenses until good computational cameras are available. After capturing your FF super shallow DOF footage, you can edit and upload from the same device, anywhere there's wireless access.
We're still using CMOS sensors- lots more to come with quantum film and similar technology. See also the effective resolving power of the eye-lens-brain system of birds of prey (e.g eagles) in a tiny package! https://allabouteyes.com/best-eyes-animal-kingdom
Mantis Shrimp are super cool too (see also their amazing 'raptorial' striking clubs): http://thescienceexplorer.com/nature/12-coolest-eyes-animal-kingdom
Chips in the body are probably not what we want- ultimately that could result in remote control (of you), so perhaps best to opt out on that one haha. Collectively we should look carefully at AI too and what we automate- the primary concern is the concentration of power of the few vs. the many. For example, I would not recommend any AI driving car that isn't 100% open source that you can inspect yourself and verify that there are no remote control back doors (or verify by a collection of trusted third parties given the complexity).
puredrifting
Major Contributor
I am 100% out of AI driven vehicles. I have been consulting for some automotive firms that are really tuned into the research and have been looking at accidental deaths with Tesla's implementation. No thanks. At least not for a few decades. My own cars primitive version of radar detect lane departure and cruise, albeit basic, has scared the crap out of me a few times where it seriously felt if I didn't apply the brakes, the car would have easily skidded into the rear of cars stopped in freeway traffic in front of me. I definitely don't trust the deployment of the technology, I have a feeling a lot of people will die from it before its reliable enough to be counted upon.
Yes, the excitement of self driving cars is already over. Now that the buzz has been out there, the news shows are now carrying footnotes that it won't be ready for another 10 years, which means it won't be ready for another 20 years. There's no way for it to work in bad weather and snow. So the panacea of self driving cars has come and gone just like 3D movies does every few decades.
The thing I could never figure out is how all these sensors, blasting radar and whatever else they are using, didn't just jam each other's signal and turn the whole thing into one big mess. It is said that the Uber car uses 64 lasers. Imagine a dozen cars around you blasting lasers in every direction. I don't know what to make of that. Or Google's radar. How healthy will be be to have every car around you blasting radar signals at you?
The thing I could never figure out is how all these sensors, blasting radar and whatever else they are using, didn't just jam each other's signal and turn the whole thing into one big mess. It is said that the Uber car uses 64 lasers. Imagine a dozen cars around you blasting lasers in every direction. I don't know what to make of that. Or Google's radar. How healthy will be be to have every car around you blasting radar signals at you?
nutmegger
Well-known member
I was shooting at the New York International Auto Show last week. Electric autonomous vehicles are the future, whether you like it or not. China is driving the auto market today.
I very much want autonomous cars. In my decrepit years, nothing will be more freeing than having a vehicle take me around when I can no longer do it myself. I was very excited about it. But the cracks are starting to show. They may be at the 95% complete stage. The question is, can they get it to the 99.9999% stage, which is where the airline industry is and where they need to get to.
puredrifting
Major Contributor
Def the future but I wouldnt ride in one today, research the gruesome deaths with the Teslas happening lately. Sure, once they sort it all out, they'll be great but I also agree, might take longer than most people think.
Around 10-15% of humans have limited or no depth perception: https://www.mediacollege.com/3d/depth-perception/ . Thus it's clear a single camera and inertial guidance system (gyro, accel) are all that's needed for sensors along with consciousness level intelligence. Even a single camera will work with no other sensors (try driving an RC car with a single camera (same as a video game / driving simulator)- in traffic you'd need more camera angles).
Currently these driving systems are inverted: many sensors and basically no intelligence beyond very trivial inference systems. NVidia, Google, IBM et all tell everyone how intelligent their AI is, and for some limited domain problems, they are indeed pretty cool (including beating humans at not only chess but Go!), but it's really equivalent to taking a trivial part of the brain and wiring it up for extremely focused and limited optimization problems.
The big problem with inference systems vs. general AI / consciousness is that any real-time input state that is sufficiently outside the training set and not part of the 'generalization set' of viable solutions can result in disastrous results when dealing with cars and airplanes etc. While beating humans at Go or Jeopardy makes a fun dog and pony show, it's not even close to toddler level human intelligence. Common sense is a vastly more advanced level of intelligence than what is currently displayed to the public.
Given advances in quantum computing, using what's been publicly revealed it's possible to extrapolate what's currently top secret which can be 10-50 years (or more!) advanced. Thus it's possible consciousness level general AI is operating on Earth right now (and for some time). That would be a very closely guarded secret on the level of zero point energy (which exists everywhere in the universe), faster than light communications (via quantum entanglement), electrogravitic propulsion (via specialized high-frequency electromagnetics, e.g. antigravity, faster than light travel), and the existence of extraterrestrial life.
Years ago there was a joke in cognitive engineering at UCSD: future jets would be controlled by a pilot, a computer, and a dog. The pilot would be present to make sure nothing went wrong with the computer and the dog would be present to bite the pilot's hand if they touched anything. This was based on most accidents being human error. It's clear now given recent crashes (cars, airplanes) that human-level intelligence with valid situational awareness (including from valid sensor input) need to be able to disable/override "AI" systems when they make mistakes.
In summary, the level of intelligence needed to safely drive a car or fly a plane, etc., is far more advanced that what is publicly available. And this same level of required intelligence is currently classified to above top secret / cosmic etc., so not available to the public.
Back to the thread topic a bit: our civilization is reaching an apex moment where we've got to make significant changes to how we all live together on this spinning spherical spaceship called Earth. The concept of "jobs" and how we live and work will change dramatically as automation eliminates more and more "jobs". Alan Watts discussed this back in the 60's and 70's!:
We've long ago reached automated prosperity level production for every single person on this planet for a basic living. This doesn't mean nobody has to work, however nobody would have to worry about basic needs for survival. Most people want to be productive, to contribute, to be valued and appreciated, so no issue with finding enough people to keep the automated systems running.
Like most things the real-world answer is somewhere in the middle; right now the system is designed to concentrate wealth and thus power for a very small population, which isn't sustainable and why there's so much suffering in the world, and it doesn't have to be that way. It's something we should all be talking about, every day, and how to efficiently, compassionately, and empathetically work towards a better way to live.
Currently these driving systems are inverted: many sensors and basically no intelligence beyond very trivial inference systems. NVidia, Google, IBM et all tell everyone how intelligent their AI is, and for some limited domain problems, they are indeed pretty cool (including beating humans at not only chess but Go!), but it's really equivalent to taking a trivial part of the brain and wiring it up for extremely focused and limited optimization problems.
The big problem with inference systems vs. general AI / consciousness is that any real-time input state that is sufficiently outside the training set and not part of the 'generalization set' of viable solutions can result in disastrous results when dealing with cars and airplanes etc. While beating humans at Go or Jeopardy makes a fun dog and pony show, it's not even close to toddler level human intelligence. Common sense is a vastly more advanced level of intelligence than what is currently displayed to the public.
Given advances in quantum computing, using what's been publicly revealed it's possible to extrapolate what's currently top secret which can be 10-50 years (or more!) advanced. Thus it's possible consciousness level general AI is operating on Earth right now (and for some time). That would be a very closely guarded secret on the level of zero point energy (which exists everywhere in the universe), faster than light communications (via quantum entanglement), electrogravitic propulsion (via specialized high-frequency electromagnetics, e.g. antigravity, faster than light travel), and the existence of extraterrestrial life.
Years ago there was a joke in cognitive engineering at UCSD: future jets would be controlled by a pilot, a computer, and a dog. The pilot would be present to make sure nothing went wrong with the computer and the dog would be present to bite the pilot's hand if they touched anything. This was based on most accidents being human error. It's clear now given recent crashes (cars, airplanes) that human-level intelligence with valid situational awareness (including from valid sensor input) need to be able to disable/override "AI" systems when they make mistakes.
In summary, the level of intelligence needed to safely drive a car or fly a plane, etc., is far more advanced that what is publicly available. And this same level of required intelligence is currently classified to above top secret / cosmic etc., so not available to the public.
Back to the thread topic a bit: our civilization is reaching an apex moment where we've got to make significant changes to how we all live together on this spinning spherical spaceship called Earth. The concept of "jobs" and how we live and work will change dramatically as automation eliminates more and more "jobs". Alan Watts discussed this back in the 60's and 70's!:
We've long ago reached automated prosperity level production for every single person on this planet for a basic living. This doesn't mean nobody has to work, however nobody would have to worry about basic needs for survival. Most people want to be productive, to contribute, to be valued and appreciated, so no issue with finding enough people to keep the automated systems running.
Like most things the real-world answer is somewhere in the middle; right now the system is designed to concentrate wealth and thus power for a very small population, which isn't sustainable and why there's so much suffering in the world, and it doesn't have to be that way. It's something we should all be talking about, every day, and how to efficiently, compassionately, and empathetically work towards a better way to live.
F4 Phantom had so many gadgets in it, the US pilot would duct-tape over them because they negatively effected the reaction time. Or, as my favorite player of all time Gerd "Der Bomber" Müller once said, "Wenn’s denkst, ist eh zu spät" (If you are thinking, you're already too late)
PS. I think you're all wrong about smart cars.
The IDF glued rear view mirrors to the sides of the F4 cockpit to provide a 'check six' solution (fixing a design limitation (solved with the F15, F16 and later).
Nice example of 'drive-by trolling' lol; meta humor for the topic
After reading, "The Mismatcher's Guide to the Galaxy" I think we can generalize the drive-by: "I think you're all wrong about everything".
On a serious note, I think AI and sensors can help make cars/vehicles safer by providing more feedback to prevent collisions, loss of control, etc. For example, recent cars provide stability systems, warning tones when signaling for lane changes, parking range, etc. The currently primitive as a binary tone (lane change) would be nice to have a visual HUD and better audio feedback for accurate range + location estimates. Breaking systems also provide more stopping force than input when range finder feedback systems determine the car isn't stopping fast enough (to prevent rear-end collisions etc.). Tesla's can also 'run away' from an impending rear end collision. Maybe not a good idea when stopped at a train crossing (or any cross traffic area).
A tricky setting is when to allow a human to override automated responses when the automation system is making a mistake. Another example is needing to turn off the collision avoidance system and airbags in the very rare case where a vehicle must collide with objects to get passengers to safety (trees, shrubs after being pushed off road in multi-car pile up, stuck in snow and needing to go through a fence/barrier (where walking could result in freezing etc.)). Fully automated cars can also be hacked or controlled by others (for any reason imaginable). Simple summary: important to be able to quickly disable automated systems by simply pressing a single button. And why it's important for the source code to be open to public scrutiny and why forcing automation with secret control code is a bad idea.
- Status