Earlier this year I wrote about what the implications of 8K video were for our cameras. You're going to start seeing some of that soon, I think. It certainly will become clearer by fall (the Tokyo Olympics were the target). Full frame at 33-46mp is going to become more the norm.
But wait, there's more.
What are the implications of 12K video?
Yes, I heard you groan. Yet 12K is certainly coming. I've heard now from three different sources about two different camera makers headed that direction. So first of all, let me reiterate this: transistor size continues to go down, and bandwidth continues to go up, and these both will change unrelentingly until we hit some real physical limits. Computers will get to those limits first—though Apple's bundling the RAM into the processor just made a dramatic change that turned heads—while cameras are actually lagging what is possible. But computers aren't there yet, so cameras aren't even close to what can be done at the moment.
Photo diodes—those are the bits of a sensor that converts photons into electron holes—don't really benefit from size reduction, thus there wasn't any urgency on the part of image sensor makers to use smaller process sizes. At least until the smartphones came along. Also, you really needed CMOS to fully take hold to get into the process reduction pipeline. Well, we've been there for awhile, and we're now regularly seeing image sensor process sizes drop (though again not nearly close to how small they could be).
Trying to do more in a smaller space means heat, and heat has implications for getting accurate image data, which we've seen with Canon's R5 trying to do state-of-the-art 8K. But to an engineer, that's just one more problem to solve, and thus, it will be solved.
8K is 33mp, while 12K is 80mp. More interestingly, 80mp full frame is 36mp APS-C (which means APS-C could do 8K when full frame is doing 12K).
One company is already at 12K: Blackmagic Design (BMD) and their Ursa Mini Pro 12K. The 80mp sensor of that camera is essentially APS-C! BMD also uses a non-Bayer array, something we're likely to see more of as pixel counts get higher and higher.
The next 18 months are going to be interesting. The camera makers are all going to take a significant step forward, every one of them. Some, like Canon, have started that step (8K) but have more steps coming. Some, like Nikon haven't yet taken that step but not only will, but are also working on a bigger step for the end of that time frame.
Finally, I should point out something I missed before all you still photographers complain that video doesn't do anything for you. It does: it provides more pixels. But it also provides something else. Actually, I had sort of noticed this but didn't actually follow through in documenting it. Did you know your Z6 II can take stills at 30 fps (eventually 60 fps)? It can, at 4K resolution. I had a small note in my books on the recent Nikon cameras that there was another set of two still image resolutions (1080P and 4K, basically) that were enabled by using the video mode, but didn't really detail what that did. Well Nikon made a couple of changes and finally got around to detailing a high-speed shooting method that relies on this. Download the PDF in the previous link and look for Split-Second Shots (where Nikon uses 1080P and 120 fps in their example, but it works in any video Frame Size/Rate, so you can also do 4K 30 fps). Focus isn't guaranteed, but for several of the things I tested this with, focus worked just fine.
Hmm. 8K at 30 fps? 12K at 24 fps? Those are interesting possibilities even for still photography.
The Tech Train travels tirelessly (say that three times, fast ;~). We're about to reach another stop.
Update: as a few have pointed out, the specific pixel count for 8K and 12K is open to some interpretation, as there are UHD, DCI, and other potential interpretations, plus if you're going to have on-sensor image stabilization, you need to account for sensor movement, too. Perhaps it's my Nikon experience showing a bit in the numbers I use here. Nikon, more so than most of the other camera companies, has a long history of pixel interpolation. In other words, they over sample or under sample at times. The D1x JPEGs were an extreme example of that, but we've seen variations on that in many Nikon cameras. Given that we don't have a three-color 4K/8K/12K solution, there will always be demosaic involved, sometimes over sampling, sometimes under sampling, and sometimes binning. So no matter what pixel count you pick for your 8K or 12K camera, there's not exactly a one-to-one relationship going on in final pixel values. At minimum, you have one-to-two in the luminance channel and interpolated color (e.g. 4:2:2 or 4:2:0).
