AFAIK a good chunk of digital camera processing is still done in ASIC/hardware and mindshift to GPU/CPU SIMD/software will take time, so it will probably be a while until camera manufacturers even begin to catch up.
That said, Olympus does a couple computational photography approaches (handheld multishot high res) and Fuji recently introduced in-camera HDR - but I tried neither.
> AFAIK a good chunk of digital camera processing is still done in ASIC/hardware
I’m not authoritative on this topic, but I can’t think of a camera produced by one of the majors which isn’t mostly or entirely ASIC based for image processing.
> Fuji recently introduced in-camera HDR
Canon has had in-camera HDR for years, but I’m not sure I’d really put the output in the same category as a modern smartphone.
They should drop their custom ASICs and just stick a Qualcomm Snapdragon as a replacement. They could use Android and create a platform with an app store with APIs to access their sensor. Opening up the hardware in this standard way to developers would really add a lot of value to their cameras.
I have no idea why they haven't done this already.
I’m sorry but I cannot fathom using Android (or iOS) on a professional camera. My Canon gear is extremely fast and responsive and Android is anathema to that. It powers on instantly and I can flip through menus as fast as I can physically turn the control wheel. I can change batteries in less than a second. My cameras will run for years without slowing down. My 7D is ten years old and as performant as the day I bought it. (It’s actually better since a firmware update lets me hotswap memory cards with my 5Dmk3.)
They can process raw images in body and display them at any zoom level without lag, loading or any other issue. My R5 has a touch screen but it’s faster and more comfortable to use the physical controls, especially since you can change settings without actually having to even look. It’s almost identical to my 5Dmk3 except they moved a couple of buttons because of the articulating screen hinge.
This is exactly what most photographers want. Familiar hardware that doesn’t get in the way. That gives the best optical quality that accurately and faithfully captures a scene giving total creative freedom to the photographer not the processor.
The idea of ruining a perfectly good system by pushing Android onto it ‘because apps’ or whatever it is such HN/disruptive nonsense.
Look at any smartphone photo on a large screen and they’re basically mush when it comes to detail. This is fine for most people, these phone cameras are far better than the point and shoot market they have rightfully destroyed.
People edit professional camera photos on large screens and nobody wants to use their camera to edit. Nobody wants their professional camera to introduce computational artefacts onto their exposure. Nobody wants to wait for it to boot. Nobody wants battery life to fall through the floor.
Lytro made a light field camera that ran Android. It was a fun gimmick and is now gathering dust on my shelf. Now Zeiss is making one for $6k and no one who actually cares about photography will buy it.
Fun fact, Android was originally designed to be a digital camera OS.
> The early intentions of the company were to develop an advanced operating system for digital cameras, and this was the basis of its pitch to investors in April 2004
The user interface in my A7ii is actually Android, mostly. Responsive Android is absolutely doable if you design it a certain way, and there is no need to remove features like removable batteries or literally anything else.
There is also absolutely no need for it to introduce artifacts into your images. RAW+JPEG is already good enough for this.
The actual advantage of an open camera are huge. For example, it would be trivial to adapt the camera to literally any mount, from EF to E to G and so on, which in and of itself is completely disruptive.
Phone processors have other aces up their sleeve. For example, unlike the Canon EOS R5, they can actually process 8K video indefinitely without breaking a sweat.
As for battery life, you're once again very wrong. The biggest draws in battery life in a camera are the processor and the sensor. Using a phone processor would allow for much lower energy usage as the lithography is incredibly more efficient, and all the high-battery-usage parts of a phone would disappear, such as always-on LTE modems, background processing, and so on.
For boot, Android with a camera sized battery and no modem can last weeks in standby, months in sleep mode. Your camera already doesn't turn off, but instead enters a low-power state. Try removing the battery of an A7 or EOS R, discharging the internal energy storage, and see how long it takes to start.
The actual advantage of this is that it allows you to become a camera manufacturer for much lower costs, meaning that you don't have to limit yourself to a single mount, and can outsource some R&D externally. For example, you won't need to be Fuji to have accurate film sims, you won't need to be Canon to have native level EF compatibility, and you'd even be able to do revolutionary things like autofocus manual lenses.
Beyond that, you'd be able to do things like temporal noise reduction, 3D depth mapping for haze removal, and so on, that can be either used for automatic processing and/or made available in RAW format for further editing.
Basically, you're describing "what if a phone company made a bad camera". What I'm talking about at least is "what if photographers made a camera untethered by the restrictions of existing cameras using commodity hardware". I think the second has the potential of seriously providing value.
> For example, it would be trivial to adapt the camera to literally any mount, from EF to E to G and so on, which in and of itself is completely disruptive.
That’s a bit hyperbolic, the collar isn’t the only difference between those lens connectors.
It is a bit hyperbolic, but both the EF, E and Nikon G mounts have been essentially fully reverse engineered as far as driving a lens. The issue is that the communication to the body of some information isn't fully understood yet and might never be, which prevents third-party lenses as well as adapters from working properly.
If you were to make a camera that is fully open, maybe even modifiable on the info it sends to the lens and what information it wants back, then you would get aftermarket adapters for Canon EF (you could also make one yourself with publicly available info). Sony E would be a tougher nut to crack, but there is a fair bit of info already out and there are already E mount adapters feasible.
So yes, maybe trivial is hyperbolic because of the Sony E mount, but very feasible.
I agree the mounts have been mostly reverse engineered, but that doesn’t detract from my point (which I might not have done a good job of articulating). The point I was trying to make is that each mount has different focal characteristics which would make it non-trivial to support on a single body (with a swappable collar adapter). Doable, yes, but not trivial. Even ignoring the focal differences, electrically (aka lens to camera communication) wouldn’t be what I would call trivial either.
(For background, a friend and I worked on designs for making an open camera platform a few years ago)
Oh, I didn't mean that the physical mounts have been mostly reverse engineered. I meant that the camera→lens communications have been fully to almost fully reverse engineered.
Mount optical characteristics are very simple. Just make sure your base mount has a larger diameter and a shorter flange distance, and you're golden. If you are designing your camera around that it is almost trivially done.
The code for implementing a basic E mount is on github, for the EF mount you just have to go looking around on some forums or buy it from the guy behind Metabones, same for Nikon G.
I have no idea why this was downvoted. Android is bloated and clunky, and far too many manufacturers put it in devices where it solves precisely zero problems and creates hundreds.
I hear your concerns, I recognize your needs. I just don't see that rebasing the camera's software atop something more flexible & open-to-building-atop like for example Android would in any way pose a threat to the camera as you know & love it.
The fact is there's no reason you'd even have to know a camera is running Android. You can put whatever user-interface shell you want on to an OS, and it could run the same form factor, present the same (hopefully better) menus and interfaces & buttons, flip through menus just as fast, have the same optics, allow the same batteries. Get Micron or someone to use persistent memory so the boot time is even less than what you have now, which is not, as you say, instant, at least not on any camera I've ever seen.
New cameras like the Sony A7S iii have really good buttons-or-touchscreen interfaces. I think a lot of people have thought buttons are the way to go, have a die-hard perspective on what a proper camera is, but they, now that the future is here, are finding the flexibility offered amazing, finding that they would never want to go back.
> The idea of ruining a perfectly good system by pushing Android onto it ‘because apps’ or whatever it is such HN/disruptive nonsense.
The software defined world is one of open possibilities. "Because apps" is such a droll unimaginative slander of that notion. It's great that you feel so well served by your fixed-function cast-in-stone device, but it A) doesn't have to be that way forever, for every-one, and B) if cameras do get more flexible & capable of user-defined behaviors, it doesn't mean you have to lose this thing you evidently love.
Allowing people the flexibility to explore additional ways of doing photography seems to me like it should be obvious & is a frontier I look forward to opening.
I come from a different place. My friends are all pro-sumer, not professionals, but we're all insulted beyond belief how inflexible, how bad the computational photography is on our fancy mirrorless cameras, especially as compared to modern Pixel cameras. More-so, failure to start allowing more creative uses of cameras is an existential risk to cameras. Yes, some photogs will keep readily buying forever, but a lot of consumers find way more value from their phones, and cameras ought to want to be able to compete, to be devices for creativity. Right now, they are, but primarily inside the same lines they have been for decades on end now.
> This is exactly what most photographers want.
Not that big a market, compared to the rest of the world, who also enjoy taking photos, it turns out. But I think the photogs would be better served too, in the long run; much better served.
Sony cameras secretly are Android devices. You can write your own Android apps for them[1].
Why this isn't advertised, promoted, & made available? Why digital cameras seem to want to get their lunch taken by phones? Not sure!
I think it was way too soon, but Samsung had a good go with a series of Samsung Galaxy Cameras[2] that were proud Android brandishing devices. Android has gotten better, mobile chips have gotten way better. But even something more like the unofficial OMF that I linked, a Sony camera, that doesn't look like an Android system, it looks like a camera, but is really Android and can run Android? That makes hella-sense to me.
Well, only the older ones, the A7iii and onward don't have an accessible Android subsystem. But even for those before the A7iii, it's more of a subsystem and doesn't have access to a lot of functions that are done by specialized chips.
This is a huge area for disruption IMO. Using a Snapdragon chip plus an FPGA/ASIC to translate SLVS-EC to MPI and buying a sensor would allow for a camera with much better performance characteristics possibly even at a lower price, and the programmable hardware would open up an insane amount of possibilities and allow for incredible value.
If anyone wants to consider doing this, shoot me a message.
They don't even need to go full smartphone processor and OS, adding something GPGPU-esque to offload the computations to would work. Deep Neural Nets brought us many "edge TPUs", I don't see why they wouldn't equally work to process images quickly and they're reprogrammable.
That said, Olympus does a couple computational photography approaches (handheld multishot high res) and Fuji recently introduced in-camera HDR - but I tried neither.