I’ve begun to think of technology advances much the same way I do about viruses: there’s constant on-going evolutionary mutations, mostly mild, with only an occasional outlier that establishes a new species and has any true new abilities.
I’m not alone in this thought. Even the BBC once said “the hi-fi world has become something of a graveyard for bright ideas that come to nothing,” but that the CD was a “real” change.
This idea of failed (tech) “species" with occasional successful new ones is something I’ve both experienced and observed in my long career. Moreover, the way in which some tech species fail varies considerably. I’ve been at companies with products that died because of cash flow issues, licensing arguments, and even the very weird “too successful to continue as a Subchapter S while unable to transition to Subchapter C due to the tax consequence.” It isn’t always lack of sales, poor/wrong product, or economic recession that kills off tech products.
One thing I admire about the Japanese engineering teams is that they're exceptionally good at producing small mutations (iterations) that inch a tech species forward. The thing that everyone admires about Silicon Valley is different: the Valley seems to be able to spontaneously combust a new hardy tech species from out of nowhere.
We need both things to happen to get better cameras. Call it in-box and out-of-box thinking if you want to use the cliched terminology.
Most of the criticism over recent camera introductions is that it's all in-box thinking. Some more pixels, some more attention to focus algorithms, faster movement of data, and so on. What I hear becoming louder in the customer complaints is requests for things that currently aren't being done in our cameras. And, of course, I've been a long time complainer about cameras not yet having really moved into the 21st century in terms of moving data over the airwaves.
One of the biggest requests from users these days has to do with computational computing. Caution: many are conflating what they see that looks okay on a smartphone screen with "computational imaging makes better photos."
Before we go further, let me correct one thought many have: our cameras already have plenty of computational computing. The complaint really has to do with the fact that smartphones seem to be doing more, and are doing it IRT (in real time). Moreover, we've had some computational stuff before that has been removed from later cameras. Remember Nikon's BSS (best shot selection)? That, or a newer mutation of that would be truly handy in a Z9 that pops off 120 frames per second on a pre-shot capture (e.g. "ignore the images of the bird just sitting, and only show me the ones of it taking off and still in frame").
We have two competing changes in final pixel approach going on right now that also need to be better resolved. The first is AI completely making up pixels; the second is computational reconstruction of pixels. In the former, new pixels appear out of nowhere. In the latter, pixels are formed by examining before, during, after pixels and putting the best pixel forward (or combining them to remove quantum shot noise). Perhaps we want both mutations, but I'd also like to have more control over both when and how each of them are used.
This is actually one place where the smartphone and dedicated camera makers can (and should) be different. Smartphones just give you their best guess. No one wants deep menus and lots of buttons to control things on their phones. Smartphones as a camera are a convenience product, not a high quality, high performance tool. Dedicated cameras (mostly) assume an intelligent controller (you) is making intelligent decisions (or overrides of decisions). The difference is important. Not everyone wants the smartphone simplification, but not everyone wants to be stepping in all the time to assert more control. These two groups of users can and should coexist. The question is whether you do both in the same device.
