"What do you think about the sky replacement feature that everyone seems to be running around touting now?"
I'm not a big fan of it, mostly because people are using it completely wrong. Basically, what you're doing when you replace sky is something called compositing. In Hollywood, where compositing mostly originated, there's great care taken about two things in the scenes being melded: perspective and lighting.
Topaz has been running an ad recently that features a sky replacement at Arches National Park, for instance. The results look completely artificial. That's because the light is hitting the rock from one direction and the clouds from another. Moreover, the lens used to take the two photos was different, so the clouds don't appear to be the proper distance from the arches.
If you're going to use sky replacement to "take photos" that didn't exist in real life, you'd best be collecting cloud shots that allow you to match lens perspective and lighting cues. Otherwise, your images will call out "fake" to anyone that lets their brain actually work at deciphering the image.
Landscape photographers—for whom sky replacement has the most appeal—have long been messing with reality. I've seen famous photographers whose name you'd recognize pull up a blooming plant from where it was and place it in a different spot in their image. In a National Park, no less (and yes, I reported this to the park service). Cloning out RVs and beer cans and a host of other "defects" is well established now. The list goes on.
It used to be that those of us in the editorial process at a number of key magazines—National Geographic being the leader—policed manipulations of all kind. At least until NatGeo moved a pyramid on their cover ;~). We had a policy at Backpacker while I was running it to label any manipulated image as "Photo rendering" or "Photo composite," and we wouldn't use such an image unless it was necessary for a point.
The Internet (and two recessions) has broken that. I don't know of anyone in the media that's doing a good job of labeling photos "photos" and manipulations "manipulations" any more. A few still give lip service to that, but I keep finding exceptions they don't label correctly.
I'm not "anti-art." I believe that a great photograph can go beyond just rendering the scene in front of someone perfectly. But I'm not a fan of "bad art." So, if you're going to do sky replacement, learn the basics of compositing and do it right.
"When companies say '5 stops CIPA' for an image stabilization method, what does that mean?"
Excellent question. If you want to know the full details of how such testing is done, you can read the CIPA stabilization standard.
Measurements are made on a standard test device at 20x the focal length (full frame equivalent). For a 50mm lens, that's 1000mm, or only 40 inches! For a long telephoto like a 500mm, that would be 10,000mm, or 33 feet.
What the standard actually measures is white/black boundaries on a standardized test chart (though, curiously, not at the angle that most MTF measurements are done; that's because they're not looking at contrast, they're trying to ascertain specific movement [change of angle]). Moreover, the white and black levels are "normalized" in processing to 0 and 100%, which may not be the way you represent tonality in your image. No aperture is specified (surprisingly), but the shutter speed is supposed to be "around" 1/focal_length ("around" is not the kind of specification you'd like to see in standards testing).
The baseline measurement (no vibration/camera movement via the special shaking platform used) is done taking at least ten images, and then the vibration testing takes ten images twenty times. This is then repeated by taking shots at reduced shutter speeds. (From a technical standpoint, your baseline and vibration results should probably use mean values for the same number of shots done the same way, but that's not true of the CIPA standard.)
From there, things get into some complex math derived from white/black transition data. But the critical element being looked at is the deviation from the "average vibration angle" without stabilization. In other words, at 1/focal_length the white/black transition had some measured value, while with the stabilization turned on it had some other measured value. The "stops CIPA" value is created by using the shutter speed difference between two "equal" shots.
So, if 1/125 had X amount of vibration without stabilization and 1/60 has the same X value with stabilization, the stabilization would be said to have 1 stop CIPA improvement. Values are rounded to half stops, so that means that two cameras saying they have 4 stop IS capability could actually vary by almost one stop!
Note further that the actual vibration used by the CIPA test devices only produces movement on two axes (yaw and pitch). The waveform (vibration) is slightly altered for light cameras (<400g) compared to heavier cameras (>600g) due to mass differences (for devices at 400-600g both vibrations are used).
All this is a long-winded way of saying "CIPA testing is no more accurate in the field as is the EPA mileage estimate for your car."
"I'm a Nikon DX DSLR user and want to go both full-frame and mirrorless. Should I get a Nikon Z or a Sony A?"
I might surprise a few of you with this answer.
The answer: it depends upon what kind of shooter you are. If you're an all-automatic-just-use-the-defaults type of user (and can get past Sony's menus for what you do want to change), then I'd tend to say the Sony (current A7 or maybe A7R). The reason for that suggestion is simple: Sony does a slightly better job handling focus when set to all-automatic shooting, though I'd advise you to tune your JPEG settings a bit.
Too many people want a camera that "mostly does the right thing." It's that mostly that gets the sophisticated shooter into trouble, but doesn't prove calamitous to the truly casual or beginner photographer. Simply put, the Sonys seem a bit more likely to figure out the right subject when you set them to "let the camera do all the control" autofocus. But they also sometimes don't always nail the focus precisely if the subject is moving fast in those all automatic modes.
If you're a more sophisticated shooter and/or like being more in direct control of things, the Nikon Z's are a better choice, in my opinion. Not only is the Z UI much like the Nikon DSLR you're used to, but the Z's also have a well thought out approach to simplifying and organizing that UI. As I've noted, in terms of command-and-control neither the Z6 or Z7 is exactly at the level of the D850, or even the D780. But what Nikon did implement in the Z's is understandable and simple to use in practice.
But the changes to the autofocus system on the Z's are just enough that you will need to relearn things in order to maximize their capability, even with the new firmware updates. I still get "but it doesn't focus as well" complaints that turn out to be a misunderstanding of something that Nikon changed in the Z cameras from the DSLRs. As I've noted many times, I have no problems shooting sports or wildlife (even BIF) with the Z's, but it took me a bit to figure out what changed and how I had to adjust what I was doing.
I could, of course, point out a ton of things that one brand does better than the other. For instance, Sony simply hasn't mastered compression in raw files; a Sony raw file with no embedded artifacts can be 2x the size of a Nikon raw file, for instance. On the flip side, Sony allows configurable buttons to be pretty much set to any menu item, while Nikon is much more restrictive (though thoughtful about that for the most part). That list of differences would go on forever, but it wouldn't help you make a choice between brands.
Indeed, there's a type of marketing and sales presentation that depends upon you being confused by all the small differences. You can almost always tell who's using that by how many people seek out "expert opinion" to "inform" their decision. Happens with cars, electronics equipment, even insurance policies.
So choose your poison well. A Nikon Z is a more natural upgrade path for a Nikon DSLR user that isn't just going to shoot on defaults all the time. A Sony A may seem better out of the box and on defaults, but can be a real head-scratcher once you have to dip into the menu system.
