Should I get a APS/DX sensor camera or a full frame/FX one?
If you have to ask, then APS/DX.
For over a decade I've been writing about DSLRs: "if you can't get a decent quality 13x19" print—the largest the desktop inkjets can generally do—then it isn't the camera that's the problem."
Many of us started getting double-truck (two-page spreads) published in magazines with digital cameras back in the days of 3mp and 4mp DSLRs that were APS/DX. Have things gotten better? Sure. But for both APS/DX and full frame/FX.
There's a difference between the two sensor sizes, sure. But if you're not understanding from the get go what that difference is, you're probably not going to take advantage of it. Save some money, some weight, some size, and get a competent APS/DX camera. At the point where you know what it is that you're missing, then consider upgrading to full frame/FX.
Same answer is true for 1" versus m4/3 or APS/DX, and for full frame/FX versus Medium Format.
Should I switch from Canon to Nikon (or Nikon to Canon)?
Most people considering this are: (1) impatient; (2) always seeking greener grass; or (3) nose down in mostly meaningless details. We're 18 years into the DSLR era. In those nearly two decades, there have been times when Nikon leapfrogged ahead of Canon, times when Canon leapfrogged ahead of Nikon. On average over time, though, we always end up with both companies getting to about the same place in terms of performance.
The biggest differences are which lenses you can mount on them and the different ergonomic philosophies. Thus, once you've bought lenses of one mount and learned one style of controlling the camera, you're basically committed. I've watched far too many folk switch and get hung up on the ergonomic differences and switch back. I've watched another group constantly selling off their entire kit to buy a different entire kit, which only remains better for a short period of time. Seems like an inefficient use of money for minimal, temporary gains.
What is the best raw converter?
The one you spend time learning all the nuances of. Assuming it's from a viable company.
This is similar to chasing the latest sensor, the latest camera, the latest anything. What any product can do is only as good as what you can make it do. Please read that sentence again and understand it. I can hand you the keys to a Porsche 919 Hybrid, but it's unlikely that you're going to win the Le Mans with it (even though professionals have done just that the last two years). You'd need a lot of training and time in the cockpit to eek out the best from a vehicle like that and keep from crashing.
The same is true of raw converters. I'll be frank. Adobe Lightroom and Adobe ACR are certainly not the best raw converters out there. That's particularly true for Nikon users. But do I use them? Yes. Because I've spent a lot of time and energy trying to learn their nuances and force them to bend to my will.
We're in another period with a plethora of choices for converting raw files. Choose one that seems like it will be around for awhile and learn the bejeebers out of it.
Caveat: Fujifilm X-Trans shooters can mostly ignore the above answer if they use a Macintosh. There is a "best" macOS converter out there: Irridient Developer. Why? Because it seems to be the only one that doesn't make a muddled mess of low level color detail from X-Trans raws. Note: Irridient now has a "transformer" version for Windows in beta that will convert X-Trans raw files into DNG files using the Irridient raw processing algorithms.
Who Makes the Best Lenses?
Same answer as to "who makes the best salad?" It varies. It might even vary with my mood today.
This is a tough question, because in optics there are so many tradeoffs that are being balanced against one another. You can press for higher sharpness, but if you try to do so across the entire frame you start grappling with things like field curvature. Every design choice tends to be a tradeoff when creating a new lens.
A lot of folk get hung up with "numbers" in measuring lenses. But frankly, there isn't a single complete set of numbers I know of that has a close and direct correlation to how good I think a lens is. And when people start running DxOMark numbers at me, I tend to just get all Trump on them. Or maybe it's anti-Trump. I just know I get angry when someone tries to make a point with the DxOMark numbers without understanding how they were generated.
Now what some people are really asking is this: who makes the best entire line of lenses? Even that is getting difficult to answer. In the film era there were clearly better brands than others. As time has progressed, and especially lately, it seems that everyone has discovered new ways to get their software tools to make better balancing decisions. Consider this:
- Canon has been consistently upgrading their lenses, and each generation is definitely better. Most weren't slouches to start with.
- Fujifilm has taken their long experience in making great lenses for things like television and medium format cameras and parlayed that into an excellent start for the XF and GF mounts.
- Leica has always made lenses that were exceptional in some way. Nothing's changed.
- Nikon has really upped their game in the recent past, and their game was already quite good.
- Olympus took what they knew from the film Zuiko era and parlayed it into excellent lenses for m4/3.
- Panasonic, with help from Leica (see above), has also been making excellent lenses for m4/3.
- Pentax was at times a clear leader in lens designs, and was selling those designs to others, though this seems to have tempered in the recent past.
- Sigma, like Nikon, seems to have really upped their game. Unlike Nikon, Sigma wasn't at as high a level when they started this new progression. But all recent Art lenses are arguably excellent.
- Sony, with some visible help from Zeiss (see below) and unseen help from places like Konica/Minolta (yes, they still make lenses), has slowly turned themselves into a credible lens maker with some excellent optics.
- Tamron (partially owned by Sony) has upped their game recently.
- Tokina (who has licensed designs from many places, including Angenieux and Pentax) doesn't make a lot of lenses, but definitely now seems to only make lenses if they can be competitive in excellence.
- Zeiss has always made lenses that were exceptional in some way. Nothing's changed.
See the problem in annointing a winner? While there might have been some better clarity 20 years ago, I'm finding that most lens designs in the last couple of years have moved the bar into a very narrow range of very high competence.
Caveat: When cost becomes a factor—as it does in kit lenses and low-end consumer lenses—variability returns and many of these lenses have some clear and sometimes frustrating issues.
How many megapixels do I need? What's the most I should get?
Enough to support your output after cropping.
If your typical output is a Facebook page, an Instagram post, a Tweet, a Web posting on your site, or even an email, you have a pretty low bar to get over in terms of image resolution. 2mp (1920x1080) would be more than enough done right. Maybe 8mp (3840x2160) if you thought someone was going to bring it up full screen on their 4K video display.
This is one of the reasons why smartphone photography took off. The phone cameras easily get over the lower bar, plus they are super convenient in terms of workflow of getting the image to those output locales (assuming you're not overwhelming the phone's memory and your data plan with tons of shooting).
I mentioned desktop inkjet printers in the first question, above, and that bar isn't particularly high, either: 5700x3900, or 22mp.
One could argue that you're future proofing yourself by buying more pixels. After all, the TV makers are talking about 8K displays, and you can always buy a bigger printer or have a lab make a really big print.
Pretty much everything from 1" compact cameras that fit in your shirt pocket (RX100) to medium format gets over all these bars, though the 1" camera may be cropping and lens/diffraction constrained if you push it to the desktop printer at maximum size. Still, I've seen some pretty nice looking prints from the 1" sensors, even at highish ISO values.
The thing that gets a lot of people into trouble is the fact that they're sloppy or lazy. They don't have the right lens or shooting position so they're cropping away more than half their pixels. They don't take the time to optimize exposure, and they underexpose enough so that it is as if they're shooting at a higher ISO value. They don't have strong skills in post processing (or they set the camera wrong for JPEGs) and the pixels they capture have issues in them. They have subject motion, camera motion, or slightly missed focus in their shots.
I ran into one person who was solving their troubles this way: they shot raw with a 36mp D800 and then in post processing downsized the image down to a 4mp JPEG. Well, yeah, that might disguise a lot of sloppy handling, minimize things like noise, while seeming to improve edge acuity. Plus since it was raw, they were correcting their exposure and camera setting errors, too.
Funny thing is, that person was following my answer: enough pixels to support their output after cropping and post processing. They just had far more to hide in their pixels and needed the downsizing to keep people from seeing it. Seemed to me they just could have upped their skill set and saved US$3000, but some people throw money at solutions.