- CCD layering & subpixel arrangements
- aperture openings & shapes
- vignetting & chromatic aberrations
- focal distance, white balance, ISO...
- shutter speed, aperture, depth of field, focus, composition...
- post-production...
The comparison to FLAC (a storage medium) is a bit flawed, vs say the actual creative instruments (guitars, pianos). The article is more like the photography equivalent of lamenting the loss of Stradivarius's secret sauce.
Long before a photo even gets to RAW, a bunch of artistic decisions are made by the photographer and the engineers who put together the camera and lens (and their firmware). You can't turn off all the processing, and even if you could, the camera lens doesn't really work the same way the human eye does. A "lifelike" photo would have a huge field of view but be blurry anywhere except the center focal point, have a huge dynamic range but limited depth of field, etc.
And then after you go from RAW to whatever renderer makes it a JPEG, a bunch more decisions and corrections are made, with or without your input as a photographer/editor.
Even if you shoot with film, there is no "calibration to reality" per se, it's all an artistic choice in terms of body, lens, film, development techniques, etc.
Cameras are not "reality capture devices", they are advanced signals processors that turn a gazillion rays of light into images that match a certain preprogrammed look (or one of several, like "natural", "vivid", "sepia", etc. -- all of those being just different parameters that tweak the algorithms).
The trope of the two overlapping discs that tells you that someone is using binoculars could be replaced with this when the director wants to emphasize some subjective aspect.
Edited for clarity.
Take for example the cliche Tower of Pisa pose: https://www.google.com/search?q=tower+of+pisa+pose
A lens which has no distortion and no aberration still produces very different effects depending on how you compose a shot.
This is reality: https://twistedsifter.com/wp-content/uploads/2012/07/picture...
So is this: https://www.boredpanda.com/blog/wp-content/uploads/2017/07/f...
And this: https://www.boredpanda.com/blog/wp-content/uploads/2017/07/f...
Here's an example: perspective works completely differently in a photo. Brain "sees" a heavily nonlinear perspective with the focused objects seem closer, but at the same time all straight lines are perceived as straight. This is impossible to represent in a photo, it has either linear or curvilinear perspective but not both simultaneously. Non-photography artists such as painters or game developers often mix different perspective planes in one work to depict this, but this is also just a rough imitation of what they really see or imagine.
Everything else works similarly. The dynamic range can't be represented in a camera+monitor system (or camera + printed media). The colors are impressions by definition, but are also skewed by illusions well known from the color theory. The motion requires you to use longer exposures, motion blur, and composition to convey in a static photo. Sense of depth, same story.
Nothing you see can be "faithfully" represented in a photo. The best you can do is to interpret the scene yourself and use crude artistic tools (tonemapping in post, composition, capture-time camera tools such as DoF etc) to hope that it will induce more or less the same impressions when viewed. Painters use extremely low-contrast media for centuries yet are able to convey the impression of a blinding sun, exploiting the human perception artifacts.
You can see the non-ideal lens as one of those capture-time tools to convey a certain impression. It might be imperfect at that, it doesn't induce the same impression in everyone, it can be imitated in post with technical precision - sure.
People who make or listen to different mixes, listen to music with their own settings they like more, abandon FLAC in favor of vinyl, people who don't care about high fidelity in general outside of archiving, just embrace this point as well.
It's the impedance mismatch between the real world - which is incredibly complex and messy - and a snapshot of that encoded as a JPEG.
Let me give a counterexample: Leica Summilux 35/1.4 (FLE), or Voigtländer Nokton 35/1.7, both with a very similar rendering, which includes a very strong field curvature wide open. Namely, the focus "plane" bends to further away spots towards the image corners. Using these lenses in the right way can create a 3D effect, and weirdly enough, you can see that even in a viewfinder with a single eye, which conceptually makes no sense, but truly is like that. Seen it using A7/A7s, tested it with other persons, they all report this effect. This rendering is so unique that even non-photographer friends eventually could tell images apart made by the Leica (if around wide open settings). In return, it doesn't have the nicest bokeh, or more like you should keep an eye on the background, but I also picked it for its harsher/"grittier" rendering with that.
Another example: Leica Summarit 90/2.4 vs. Zeiss Loxia 85/2.4. On my travels I use the Leica, but otherwise I try to grab the Zeiss. Thing is, the Leica, like its other 90mm brethren, suffers from veiling flare (yes, even the hailed Summicron, and also the old Tele-Elmarit), a problem the Zeiss doesn't have. In return, the Zeiss is bulkier and has almost twice the weight, so for travels (multi-day hiking etc.) the Leica has advantages over the Zeiss, despite being the inferior lens from a rendering point of view. Read: Sometimes you have to compromise on "image quality" (very fuzzy term really) for other factors like usability and so on. A common theme with the more perfect lenses: Larger + heavier, and that is not always a good compromise to take.
Which means there are many factors which play into lens choice, as usual, driven by a wide range of preferences and requirements. That also means it is not really possible to make general statements for one or the other, but "depends".
* In the center of the image things behind the sujet are strongest out of focus.
* With the focus "plane" wandering off to the rear towards the corners, you can put up some "visual pillars" on the sides which are somewhat in focus, but shouldn't be from our own visual training. That seems to mentally push whatever is actually in focus in the center even further to the front. It also means if you just isolate the sujet without that most of the effect goes poof.
* On my A7s, through that field curvature, the corners are also somewhat smeared.
* The lens actually does have a lot of microcontrast in the center, i.e. is really sharp.
From what I can find online, some good examples of the effect are shown here: https://phillipreeve.net/blog/voigtlander-ultron-11-7-35-mm-...
I mentioned this lens, because it is in certain ways superior than the Leica, but in overall has very similar visual qualities. It doesn't show in every wide open picture, it requires this "context" to work its magic, and some of the examples show that.
Another lens with this effect is the Voigtländer 21/1.8, but because it suffers in general in the sharpness department wide open it is less pronounced. I think on the above linked website there should be some examples for that, too. More perfect lenses look ... different. They certainly work better if you don't want these "visual pillars", so it is not like that field curvature is always the perfect thing to have, quite the contrary. However, it is a tool, or can be used as such, like so many other things in photography.
I didn't know about all this at the time I got the Leica, only that it had field curvature wide open, and learned these things through using it. But considering available pictures online, there are maybe ways to figure the field curvature out, because most test/review sites are rather useless in that regard:
https://www.dpreview.com/opinion/4042117089/roger-cicala-why...
and for some additional info on top of that
https://www.dpreview.com/articles/1351719699/roger-cicala-fi...
Once you have seen these "special" images and their look it also makes it easier skimming through images from other lenses if they show that here and there.
Roger also jokes about the 3D-pop at the end of the 2nd link, but considering that tilt-shift lenses can produce a miniature look by meddling with the "field curvature", getting specific other looks through usage of actual field curvature isn't too far fetched of a concept. It is some sort of an optical illusion in the end.
Zoom with your feet, then. (That is, go there and look.)
With digital, what you get is determined by the compromises made by the sensor designers, or (increasingly!) by software; with analogue, what you get is constrained by your choice of film.