Today's level of integration
What makes today different from any other time is the level of integration in today's electronics. Here's what I mean.
Let's say that you want to design a cinema-type camera. Which means it has to have a resolution of around 4K, must have at least 13 stops, and be compatible with most modern (and ancient) lens-mount standards. And it has to have a neat, functional industrial design with a capable, flexible, connected operating system. The UI needs to look good as well.
That's a lot of work if you intend to start from stretch, but the point is you don't have to start there: you actually start about 98% of the way down the track.
You see, most of this stuff is available off the shelf. All you have to do is design the look of the product, write some "glue" software, and you're up and running. Don't believe me? Well, this is how it would go.
First, find a sensor. Aptina have recently produced a great-looking 4K sensor that's compatible with standard electronics. You might think that Canon and Sony have the lead in this type of thing because they make their own sensors, but that doesn't stop other companies specialising in ONLY sensors (which is what Aptina does) and because they specialise, they're actually pretty good. (You'd probably want to define anyone that supplies Nikon with sensors as "Pretty Good").
Then add some storage. There are lots of choices, with many options fast enough to store 4K RAW video.
Glue electronics - yes, complicated, but well understood, and you can fit it into a FPGA - which is software-configurable hardware. Essentially whatever it is you've built wakes up and thinks it's a toaster, but as soon as it loads up its firmware, it can behave like almost any type of logic-based processor. Programming FPGAs is no trivial thing, but there are quite a few experts around now so you would rarely have to start from scratch, and FPGA vendors like Altera build in already-working modules to deal with SDI, HDMI etc.
An operating system and UI? Well, for a modern camera you'd need plenty of processing, wireless connectivity, network connectivity, a touchscreen, and probably GPS and accelerometers as well. So you've be able to build a connected camera with a modern interface that would know where it is and quite possibly have some type of image stabilisation based on the output from the built-in accelerometer and gyroscope.
All of this might sound like a lot of work, but it isn't, really. You can buy Android boards that have virtually all of these capabilities (because, let's face it; smartphones do), and all you have to do is essentially write an Android App that is your User Interface. It would probably be pretty stable too, as people tend to update the software on their cameras less often than smartphone users. If you don't want to use Android there are plenty of other options, including the Raspberry Pi, a Linux computer that costs much less than a tank full of petrol.