You can definitely get much better images if you have a good lens, and on older cameras good lighting or very good form. But more modern smartphone lenses and software are plenty good for certain types of photography (and especially video).
I shoot less and less with my D750 and D700 (in non-studio / non-low-light-event situations), and more and more with my iPhone, because of the convenience and portability.
But there are definitely types of pictures that can't be taken without good glass and a larger sensor size. No software can simulate the physics that makes a longer focal length make a great portrait. (You just can't stuff that quality of glass into such a small lens, and the sensor can't physically gather enough photons to compete.)
Today I travel with an iPhone 11 Pro and a Fuji X-T2. In the past it's been whatever current-gen iPhone and pocket cameras like the Sony RX-100Mk3, Canon S110, Canon S95, and DSLRs like a Nikon D3000 and D50.
It used to be, the photos looked better on the phone after shooting them, just because the screen was so much better, but once imported to the computer, the iPhone photos looked like an oil painting even compared to a upper midrange P+S pocket camera, let alone a DSLR. It wasn't even close.
These days, between the XT-2 and the iPhone 11 Pro, it's hard to tell which is which. Sometimes the iPhone looks better as-shot because of the image processing presets. Other times it's the Fuji. At 1:1 on a 5k display it's just not as obvious as it was when the pixels were blown up more.
Even in good light, yes, the sharpness on the Fuji is better, but not by leaps and bounds. Not in a way that would be immediately obvious to anyone walking by the computer as it used to be.
I've never bought high-end glass, and frankly, some of my earlier DSLR photos are pretty bad compared to what my iPhone returns these days.
Seriously, even a beat-up one from eBay, just try! Costs way less than a new flagship phone and the difference is still amazing. And nowadays it's not difficult at all to learn how to use them decently, don't be afraid!
Which is perfectly fine for everyone of course, but given that you have to buy high-end (my very own 150$ Android One device has no such camera) and a central theme of this high-end innovation is that you buy a new device every 2 years (nowadays mainly advertised through the camera) and then throw the old one away, I just decided to buy a small ILC (EOS M10 on sale) and use that for taking pictures until the shutter dies (at times where my main camera is too big (and thus most would use a mobile)). Works fine for me and with 12mm/2 and 22mm/2 lenses I can (the 2010ish sensor notwithstanding) still only grin at the nightmode of the newest iPhone or the star mode of Google...
Would love to know if anyone here knows more specifics firsthand, but that camera's sharpness has always impressed me compared to the replacements I've tried out.
Some back of the envelope math: 20/20 vision = the ability to resolve differences 1 arc minute ≈ 290 microradians across. A distance of 1/300 inch subtends an angle of about 280 microradians at 12 inches, so I estimate someone with 20/20 vision could see pixelation on a 300 dpi image inside 12 inches.
Is that too weird? Well, maybe if you're hanging it on the wall, but it seems odd to put a 7x10 image on the wall unless it's in a collection or montage. I think photos printed at 7x10 are more likely to be held in the hands, in which case shooting for transparent at 12 inches seems like the way to go, actually. (I suspect many people would have trouble discerning details in a 7x10 image from several feet away.)
Photographic continuous tone images never have that sharp of contrast, though, and, again, depending on the image, you might not see jaggies even at 100dpi.
300dpi is sufficiently overkill to be an easy "set it and forget" resolution to recommend. (Source: I paid for college doing typesetting, graphic design, and digital prepress work).
I used to think 300dpi was "bare minimum", having studied computer graphics in college and Nyquist limits and all that. And then one day my father printed a 640x480 cellphone camera photo as an 8x6 and... it looked fine. I was blown away.
And then I finally remembered: human eyes don't detect absolute light value, we can only see relative differences in light values. Gradients don't provide enough change to see the edges between colors. You can see it on a computer screen because the screen amplifies the contrast of the image (another old lesson I eventually had to relearn so as to compensate when making prints of digital photos).
Nobody knows what resolution you print things when they see the print. If the average joe "knew" how the photo was made, they might scoff at it. I guess people think that craft requires fine precision at all steps of every process. Certain craft does, but some other craft requires randomization and binning, and others have diminishing returns on extra effort. So I find it kind of fun to print low-resolution photos. Feels like I'm winning a game against overly officious critics.
> Nobody knows what resolution you print things when they see the print.
Well sure, you're right about the gradient thing, but what if you have an image with some really sharp, contrasted edges?
Just anecdata, but my family and I were walking through a seasside town a while back and ended up in an artist's shop. Took a look at some photographic prints on the wall (going for $50-$150+, as "artisan" stuff often does), and I had to call my partner over to look at them because I couldn't believe they were all pixelated! Some of them quite badly. I was about 18 inches away from them, I guess, and they looked to have been printed at about 100 dpi. Sure, you probably don't need 300 DPI for large wall prints, but I'd rather have a margin of error and not have that happen to one of my prints. Most people aren't likely to notice flaws in printed images, and are even less likely to tell the owner if they do.
I guess it's possible the shots sent to the printer were badly compressed JPEGs, but that seems unlikely.
Print engines can apply nearest-neighbor interpolation which results in visible, blocky pixels. If they'd upsampled the image with lanczos or another higher-quality algorithm before sending it to the printer, it would have looked like it was just a touch out of focus.
The first image is a splotchy mess, while the second image retains fine details, only the outside edges are blurry due to the toy lens I used. The other big differences that people are quick to notice, depth-of-field and focal length.
[1] https://www.naut.ca/dropbox/IMG_1827.JPG
[2] https://www.naut.ca/blog/content/images/2019/09/P1040326.jpg
