I've done TR and Zwift ramp tests, the Sufferfest 4DP test, and had Garmin autocalculate FTP. The Garmin estimate, as far as I can tell, just took my best 20 min avg power during a longer ride, so I consider it much more arbitrary.
Every test I've taken has agreed within single digit watts on my FTP when they were taken around the same time (only training for a few years so I'm still improving my FTP) and the subsequent zones have felt right afterward. Interestingly the Sufferfest said my best relative power type was 5-min, so I am at least somewhat more anerobic-inclined, but that didn't result in an overestimate of my threshold, since I find long duration sweetspot intervals tough but doable, whereas when I'm above FTP all physiological signs show that I'm going into the red.
Before I got a smart trainer, I did all my indoor tests on rollers using magnetic resistance and a fork stand. At sub 300W I could still manage at 100 rpm. It was annoying to try to hit the power targets, so I thought I was wasting effort and not hitting the best numbers possible, but switching to a Kickr in Erg mode didn't lead to any big improvement.
I haven't tried the Zwift ramp test lite version, but my wife did one using a free Zwift trial and the results looked reasonably close to what I'd estimate from watching her training rides.