So… are people using this? I have to say that I’ve only recently upgraded to 2022.3 and I was very excited to try out the new camera analysis, but in practice I haven’t actually gotten very good results. In fact, to be plain about it, all of my solves are complete garbage so far.
For sanity, I took the same shot and ran it through the old Mono Analyzer and I got a pretty good solve almost right away. I’ve watched Grant’s awesome videos but so far no good… Is it just me?
It can be a bit hit-or-miss. Both Will and Grant have been pretty up front about that in some of the logic live episodes when they first introduced it. I have used it multiple times with success. The ML aspects of it have come through well for me, such as recognizing moving people and removing them from the calculations. Other shots that I would have expected stellar results from were complete failures.
For sure. Like anything, I expect some hit-and-miss. I’m just a bit surprised that so far I am batting .000 with it. I wouldn’t want this to turn into a SLAM-Autodesk (pun intended) session… I generally assume when something doesn’t work for me that I must be doing something wrong. I’m going to continue to feed batter in to the machine and we’ll see what comes out. 
yeah, I find myself sticking with the old one for most shots. The ability to prune tracks is a big part of my workflow and I don’t find it working as well in the new one.
But as @ytf says, it can get good results for some otherwise brutal tracks.
It’s still option B for me, or option C if I’ve got a Syntheyes license.