Camera software is an area that is frustratingly closed off, and where manufacturers regularly differentiate new/more expensive cameras with firmware differences: subject detection, auto focus, drive modes (pre-capture, higher frame rates, etc), file format support (HEIC, compressed raw, colorspaces), ui features (how custom modes work, menu systems, etc).
Competition in this space driven by a provider that doesn't have a motive to convince users to buy the next new camera (ie: open source software) would be very useful to users.
Sony and Panasonic cameras (and perhaps other manufacturers) are running Linux (and release some of the third party source code they include in their products, but the amount of reverse engineering that has been done on those camera's software.
Canon uses a custom RTOS (DryOS), and lots of work has been done to extend the existing Canon firmware to enhance its capabilities (Magic Lantern[1] and CHDK[2]), but as far as I know fully open software for these devices doesn't exist.
Camera capabilities these days are highly software dependent, and the functionality of dedicated cameras is held back by subpar software development practices and a lack of pressure to make things better.
>Camera software is an area that is frustratingly closed off, and where manufacturers regularly differentiate new/more expensive cameras with firmware differences:[...]
Why is this frustrating? How is this different than anything else tech related that's in the business of making money?
Every company trims the features they give their customers access to depending on the amount they're willing to pay.
Camera companies aren't charities, they need to make money for their shareholders. All that sunk R&D has to be recouped.
This isn't just market segmentation. It's a frustrating, recurring theme among Japanese hardware manufacturers where great products are negatively impacted or outright hamstrung by shit software support. I run into it in embedded firmware and industrial controls all the time. I think it's a side effect of the corporate culture there, which is how it manages to cut into multiple market segments. Having to pass on good offerings because the software is too bad to stomach feels bad.
I get what your saying but how is their artificial firmware limitations different than Apple's own market segmentation practices? Or Google's? Or Microsoft's? Or Hashicorp? Or Docker?
It isn't different, but it's also not a clear comparison. Some differences are based on binning and calibration for example. Take sensors, silicon, power systems etc. which are all imperfect. Some product variation exists on purpose, while others are mostly just as a side-effect from the production yields.
Some segmentation is a bit extreme (Windows Pro etc. can use gpedit but Windows non-Pro can't...), while others are a matter of paying for a service rather than a thing. (i.e. Terraform FOSS locally vs. Terraform Enterprise in the cloud - you actually get a different product with different capabilities, not the same product with different things 'cut off')
Perhaps Intel is a good example in the hardware realm; some chipset and ECC variations are choices that are purely made so that server usage is highly discouraged. I would guess that some of that is simply a side-effect of R&D cost: you can't really have consumers foot the bill for extreme R&D, but you also can't make better chips without a very costly R&D process. So they might have selected a segment they know they can squeeze and put the specific features only in products for that segment. (One could argue that another perspective might be 'server' CPUs subsidising 'consumer' CPUs)
It's not always as simple as 'fake barriers for no reason', but because sometimes it really is, it gets hard to distinguish with is really annoying.
I think the frustrating thing is they can (potentially) afford much less to fund a competent software team, so the end product is much less than it could be. Which is unfortunate because of how good the hardware is.
It's frustrating because it's economically inefficient. It's also probably bad for the camera company's shareholders. It may be good for certain company employees - what economists call a "principal/agent problem".
Or perhaps the companies think they are not in a competitive market, and so can make more money by engaging in "price discrimination", where they try to charge customers more if they have more money (indirectly, via artificial software limitations, which they think has this effect). This seems unlikely to me, though. Customer "lock in" to one company's products must be pretty short-term nowadays.
If they treated cameras as a competitive market, they would be happy for software to be developed by others, if that's not their competitive advantage (and clearly it's not). They would make money from selling the hardware (cameras, lenses, other accessories). If they're the only company with open source software, they should have a big advantage when selling hardware, and could charge more for it.
It's frustrating because much better do devices that do things that no available device can do could be made available if was more open. It's different to say, computers, where Linux is available and can be customised to your heart's desire. It's not different to lots of other areas. Those are also frustrating.
It is frustrating when any company does it! Artificially software limiting features reduces the utility of devices across the board, which sees those devices going in to landfills sooner. All of the energy that goes in to mining, transportation, and manufacturing of the materials is less effectively utilized, so these practices lead to higher emissions, more waste, and more rapid environmental degradation. They also contribute to stratification of our economy, as users have to pay more money more often to get the features they need, raising shareholder profits off the backs of consumers.
This idea that we can manufacture something complex and then artificially limit its features to run profit seeking schemes is one that makes sense in a pure business sense, but in the end is often harmful to society. We don't have to do this. We could have a fully functioning free market economy even in a world where business norms involved open source software and open source hardware. Businesses would charge more for hardware but it would provide value to customers for so much longer, that down market customers could just buy used and get the same features they would have gotten if they bought a new, feature limited device from a closed source vendor.
"Then people would rip them off!"
No, then people would build off of their work and compete on other things like price or other features. We can see as a case study the world of open source 3D printers, where few vendors sell direct clones of other models, as they tend to use other manufacturing methods and change the design for their factory, adding differentiating features along the way.
"Then no one would invest!"
No, instead of less frequent large investments, you would see more frequent small investments - investors helping a factory buy one more machine to meet production goals for the next quarter with a new design no one has yet seen. First mover advantage is real.
Notably, even the libertarians at the Mises Institute oppose intellectual property restrictions, which are an artificial government monopoly on information, an extreme restriction on free markets!
OP's last sentence explains exactly why it's frustrating - because cameras are held back by subpar software. Why do you think being able to explain the root cause of this would make it any less so?
Magic Lantern is an awesome example of the value which can be created by collaborative open source, despite the hardware manufacturer doing nothing to document or help. It's great because it's also very low risk for the user since the firmware mods are loaded on every boot, the camera is never perma-modified reducing the chances of bricking the device.
I wish one of the major mirrorless camera manufacturers would decide to openly endorse an open source model as an optional alternative to their own factory firmware.
It's not just that they want you to buy new cameras (of course they do), but they also do it to protect different verticals within their company. If a stills camera can shoot video of the same quality but much lower price, then that takes sales from the provideo cameras.
That's a great point, and shows up in annoying ways in Sony's camera line up in whether "high frame rate" support exists. For Sony cameras, HFR implies "can capture a limited length (seconds, typically) of footage at a higher frame rate (240fps, 480fps, 960fps, etc) and reduced resolution as frame rate increases (this is not the same as normal video modes and S&Q video modes, which don't have time limits like these).
Sony includes this feature on cameras without interchangeable lenses (ie: "point and shoots"), and on their high end cinema cameras, but not on their mirrorless (interchangeable lens, but not considered cinema cameras) line.
Seems pretty clear this is designed to avoid their mirrorless lineup have a feature that might make it competitive with their cinema line. But they're ok including it on the point and shoot line because that already is well differentiated from the cinema camera line.
I've always given benefit of the doubt that the limited duration of S&Q modes in stills cameras is related to the max throughput of the recording media. 120+fps at 4K resolution starts demanding quite a bit of sustainable bandwidth. Shooting to SD cards might be a limitation and explains why the pro-video cameras have much more expensive record formats.
Is there any chance that some of this limiting has legitimate hardware reasons?
It's been over a decade since I really followed the digital camera scene.
At the time, DSLRs were just beginning to get 1080p/720p video capabilities. There were frustrating time limits of course, even then. But, I recall that some of the cameras heated up pretty significantly when capturing video. This was sometimes cited as a reason for such time limitations on video recording.
I'm not sure if that is possibly a "legitimate" physical reason for the limitation today, or if it ever was in the first place. Any thoughts on that?
Don't forget about the old 4GB filesize limit of FAT formatted devices from those early cameras. That was the hard limit not time. They would say time because they knew the bit rate and how long 4GB was in runtime. It was fancy cameras that would hit 4GB, and create a new file. It was fancier editors that would recognize that it was meant to be one clip and import it that way. It was the less sophisticated that would force you to "stitch" it manually or work with multiple clips.
It still is, and I think its legitimate still - the R5 can record 8K video but of course real time encoding of that is going to generate a lot of heat.
A Cinema body has a lot more room for heat sinks, and can more easily accommodate fans - you're a lot more likely to be using an off-body mic on the cinema body. And then you have a larger battery for the chips and recording time.
>A Cinema body has a lot more room for heat sinks, and can more easily accommodate fans
And yet, you'll still see cameras on hot sets with bags of ice around them because these heat sinks and fans cannot keep up. The original Red One was notorius for overheating. I have BMD Ursa Mini that was struggling in 105 degree temps. Big body does not mean over heating is solved. Personally, the camera that I've used that best handled heat was Sony's F55. Shooting interior of an old van with sun beating down through a window in the roof directly on the camera, 110 degrees inside the van (yes it was miserable), and the camera never once complained (unlike the cast&crew).
Hope at some point we can have un-crippled firmware for the 1D(X) series of cameras. Magic Lantern developers are acting like scared cats because allegedly someone at Canon said 'No'[1] and no one questioned it.
In theory developing third-party software for devices is legal if the 'finished product' does not contain material copyrighted by the original vendor.
Competition in this space driven by a provider that doesn't have a motive to convince users to buy the next new camera (ie: open source software) would be very useful to users.
Sony and Panasonic cameras (and perhaps other manufacturers) are running Linux (and release some of the third party source code they include in their products, but the amount of reverse engineering that has been done on those camera's software.
Canon uses a custom RTOS (DryOS), and lots of work has been done to extend the existing Canon firmware to enhance its capabilities (Magic Lantern[1] and CHDK[2]), but as far as I know fully open software for these devices doesn't exist.
Camera capabilities these days are highly software dependent, and the functionality of dedicated cameras is held back by subpar software development practices and a lack of pressure to make things better.
1: https://magiclantern.fm/ 2: https://chdk.fandom.com/wiki/CHDK