With the $15,000 i've spent on gear (just added it up the other day for an insurance quote), i'd be saying:
- Yes, my images have improved
- No, my images haven't improved enough to make back $15,000 worth of print sales.
But there's a difference to "Quality of Images" and "Variety of Images".
When I first got my kit, it was a 7D, EFs 15-85, 70-300 non-L.
Good kit to start with, I got some great landscapes, I got some decent-ish wildlife.
Then add in a few CPLs. Not much, maybe $100. Quality of landscapes increased a lot.
But still, I couldn't take anything good indoors. So i got a nifty-fifty.
Shooting at iso3200, 1/30s, f/1.8, and pushing a bit in post, (it was really dark when I was using it), the images weren't great, but compared to using the EFs 15-85 at 50mm f/5 or so, I'd have been at maybe 1-second exposures. So in that way, after buying the niftyfifty, images got a lot better compared to what would have been possible previously.
Then I started reading about a thing called a Lensbaby, and bought a Double Glass Muse. Now, images certainly weren't any "better" than I could get with a niftyfifty, but they were "different", it's an effect not easily emulated (i hate PP-effects except rotation, sharpening, contrast, sat).
Then the 70-300L got announced, and I bought one a few months later, and sold the 70-300 non-L to my sister. Did the images come out better than the non-L? Damn straight they did. I haven't sold any, but I'm definitely of the opinion that something at 300mm would be acceptable to a buyer, but definitely wouldn't be with the non-L.
Further expanding possibilities, I got a few Pentacon Six lenses with a Tilt adapter. Did they produce better images than my 15-85 and 70-300L? Probably not, they were mostly uncoated, all max f/2.8 or 3.5, usable in the same aperture ranges as my zooms. But they added to the "different" abilities, namely tilting (i've also since got a shift-adapter, i'm too tight to just buy a tilt&shift adapter from zoerk or mirex).
Add in more fast primes as time goes on, 35/1.4, 50/1.4, 55/1.2, 85/1.8, 100/2.0. They may not produce "better" images than the first zooms, but when used in certain situations (like bugger-all light), they're better than what I could have done with what I had before (with slow apertures).
And a Sigma 8-16mm. Sure, it's great. But it didn't replace anything, it's simply adding to the focal length available to me (I could probably get "better" images stitching from a longer lens, but that's annoying and time-consuming).
And lately, an EOS 3, a Pentacon Six body, a Kiev 60 body (all for film, if you've never heard of them). Do they produce better images? I doubt it. Maybe the P6 with some Efke 25 scanned to 46MP might look good. But I could probably get better images from the 7D still.
There's one concept I'm skirting around, and in economics we call it "utility". Think of it like putting a dollar value on 'fun'. I've certainly had my $15,000 worth of utility from my kit, even if I never reclaim anything from selling any lenses or prints.
Nowadays, there's not much I *can't* take a photo of, be it wide angle, long distance, low-light, long-exposures (ie with ND filters), tilt, shift, macro, action, you name it. So my lens-buying habit has sort of died-down lately, there's not much else to buy to expand possibilities. So for me, there's only one thing left to do, is ditch the crappier lenses and buy "better" ones. And that's something I haven't started yet, because for me, the fun is in just being *able* to take the photos, not how *good* they print (because, well, i rarely print)