Cameras and Photography Explained

"Marry your lenses, date your body." --dpreview forum post

I wish I had come up with this response, because it's dead on (though it promotes bigamy ;~). All I have to do is look in my gear closet to see that it's true: plenty of lenses of that have lasted me years, but bodies, not so much. 

As I've written many times, I'm a fan of optimal data. That's why bodies churn through my gear: we've had serious progression on what that means pretty much throughout the digital era. Pixel counts, dynamic range, fixed pattern noise, basically all the low-level crud that can get in the way of optimal data has all been bettered over time. On average, I'd say we have been on an every-three-year rule, though it appears to now have stretched to closer to an every-five-year rule. (What do I mean by that? Every three years I was seeing clear and useful change at the image sensor in terms of optimal data, though this currently seems to be closer to five years in the latest round of cameras.)

Lenses, on the other hand, fall into one of two categories for me: look or optimal. There's no doubt that the look of a few legacy lenses was not just unique, but quite desirable (the bokeh of the original 58mm NOCT, for instance). On the other hand, starting with the exotics, but extending across the optic lines starting in the teens, we also received lenses that resolve detail clearly, and with little or no problematic artifacts. So much so that I was reluctant to part with my old 400mm f/2.8G until I could verify that the latest 400mm f/2.8 S easily matched it, but was smaller and lighter, plus it had that the built-in teleconverter that truly made it more flexible. My old 400mm lasted from D3 to D6; hopefully the new one will last from Z9 to Z9 Mark III. 

"Once sensor read-out gets to 1/400, no one will put a mechanical shutter in a camera." -- dpreview post

I certainly hope not, but I fear that this poster is mostly correct (the actual number is probably closer to 1/250 given the Z9). 

The reason is simple: the moving mechanical shutter blinds are in front of the focus plane by enough distance that they provide a masking effect on some forms of timing interactions, particularly things like LED display walls. 

What we really need before moving to all electronic shutters in all cameras is a really good global shutter sensor, and it seems that we're still years away from a global shutter that doesn't take us backwards in terms of image noise.

"Making sensor sizes less misleading." -- headline at several photographic publications

I'm sure glad they used the word misleading, because the topic is still confusing to most. 

Changing 1"-type sensor to Type 1 sensor does help, I suppose, but you'll note that unless the actual size is included, as in Type 1 (12.7 x 9.5) sensor as proposed by dpreview, most people wouldn't understand the difference between, say, Type 1/2.3 and Type 1/1.28. But the extra size info starts to become a laborious construct if you have to repeat it every time you try to identify a sensor size.

What I also don't understand is this: why you'd conform the smaller sensors, which all start with Type 1xxx, but not the larger sensors. dpreview, for example, is still going to use Four Thirds, APS-C, and Full Frame for those, despite the fact that there are three distinct APS-C sizes (;~). Meanwhile, for some reason Medium Format is now "Medium format Type 3.4". Huh? Shouldn't that be Type 3.4 (Medium Format)  or maybe Type 3.4 (44 x33mm) to be consistent?

(Sony Semiconductor doesn't help much, as they currently identify the primary ILC sensor sizes as 4/3rds, APS, and 35mm, but all the other sensor sizes as Type #.)

Let's put things into a table for clarity:

Okay, does that now makes things "clear"? From a long-time editor point of view (now 40+ years with national publications), I'd say no. 

In terms of what you need to consider when thinking about sensor sizes, I'd say only two attributes really tell you anything meaningful: sensor area and aspect ratio. Area is important because it tells you how much light can reach the image sensor. Because current sensor technology is very good in terms of capturing the randomness of photons, the area of capture is now the best indicator of potential noise in the data (all else equal). 

Aspect ratio is important to some, as it dictates whether you're going to use all the pixels on the sensor or not. For instance, I don't like the 4:3 aspect ratio. I almost always crop my m4/3 work to either 3:2 or 16:9, which means I'm not using all of the image sensor. That's equivalent to using a smaller image sensor, because I'm not capturing the same about of light. Hmm, Type 4/3 (17.4 x 11.5mm) anyone?

Basically, everyone is in pursuit of Goldilocks-type satisfaction. With ten common image sensor sizes (four are used currently for interchangeable lens cameras), we have a lot to consider. If you haven't read my article on sensor size, I suggest you do so. 

Nothing's going to really change unless the Japanese electronics companies come to a consensus standard. But that's what's given us the now outdated DCF standard and the confusing card speed standards, so maybe not...

"Legacy controls..." -- oft-used term, but often misused

I keep seeing this construct used in reference to various different brands, but most notably Fujifilm and Nikon. 

Let's start with Fujifilm: they have no real legacy in the sense that Internet posters mean. Seriously. The SPro1 through SPro5 were based on different Nikon bodies, and used varying Fujifilm kludges on Nikon controls, which differed from camera to camera in the Fujifilm DSLR era. The X-Pro1 established yet another set of controls in the mirrorless era that were effectively film-SLR-like dials, and we've since seen Fujifilm fork and modify the controls so that they now have "dials" cameras and "button" cameras. What exactly is the "legacy"? Inconsistency?

Sony can't really claim much legacy, either. Minolta controls morphed to Sony Alpha controls morphed to NEX controls (which had several variations), which morphed to yet newer Alpha controls. I didn't consider Sony's UX to have completely settled down until the recent A1-inspired consolidation, which we now also see in the A7 Mark IV (and, I hope, all future Alphas).

Canon has been all over the place. After establishing DSLR EOS cameras with a fairly consistent standard, we then got the M and R oddities, with new controls (including some very hidden from view customization on the M5/M6). I thought this was going to settle out with the newfound R3, R5, and R6 consistency, but then the R7 came along with its big dial around the thumb stick, which changes hand position on the EOS R design for no apparent reason. 

Nikon probably has the deepest control legacy, in that they established a button+dial interface back with the N8008 film SLR decades ago, and have been refining that notion ever since. Of all the brands, Nikon probably has had a bit more consistency in their controls, though they do have the annoying habit of moving buttons around. (Of course, they, too, have deviated once in awhile, most notably the N70 and the Nikon 1 models, but those were quickly relegated to history.)

I'd argue that "legacy" is mostly a mis-remembered construct in many posters' minds. They think they remember something that was consistent, that they fully understood, and which they 100% liked. Funny thing is, memory is not all that it's cracked up to be. Our brains have this interesting ability to forget pain (both literal pain and figurative pain). This tends to "improve" the past at the expense of accuracy. 

I might tend to argue that our cameras tend to have too much "legacy" in their designs and not enough embracing of new ideas. It took a long time for touchscreens to make their way to digital cameras, for instance, but I find that addition to be quite useful and needing to be explored even more. An example: while chimping my images on a D6 or Z9 I can now flick up to send via FTP, flick down to mark for deletion. With one finger I can move through the images quickly and get them rated and sent as needed. Much faster than button+dial. Which makes me wonder what else can be achieved faster and better.

Camera makers have done something that tends to be problematic in technology: they are looking rearward more than they're getting in front of us. That's one of the reasons why the smartphone photo capabilities improved and expanded rapidly and cut into the camera business in the first place. 

I'm not advocating for a complete replacement of how we control our cameras. Not at all. The way I would look at camera design these days would be to refine the base UX while extended it using technologies that are at the leading edge of what can be done. One thing I haven't seen done yet, for instance, has to do with "setting before" versus "setting after". What do I mean by this? 

Right now we must set things prior to taking an image. Aperture, shutter speed, ISO, and a huge host of other functions. What if instead we set only a few critical things and the camera could come back and say "from the data captured I see the following other options for your image"? Active D-Lighting is a really good example of this: you have to set it prior to taking the image and then evaluate whether you had the right setting. If not, change that setting and...oops, the photographic opportunity is gone or has changed. Why can't the camera just take some image data and show you the Active D-Lighting options after? Yes, that would use some buffer to keep the raw data around, and it might require the camera take multiple images quickly (exposure is changed with Active D-Lighting settings, which is one reason why Nikon thinks you have to set it before the image is taken). 

Nikon once had some capability similar to what I'm suggesting: BSS (best shot selection). The camera would take a burst of images, and then promote the one it thought came out "best" (typically sharpest). You could select one (or all) of the others, but that evaluation came after you took the image.