Just saw your posts and decided to take it for a spin. It does look to be good with props that are flat but once it goes from 2.5D to 3D it doesn't do well. Here are some I put together.
View attachment 1924764
"Salacious Crumb"
View attachment 1924767
"Stormtrooper"
View attachment 1924768
"Nien Nunb"
Although good for trinkets, I don't expect this method to ever have the precision required for accurate replicas. There is too much guesswork involved; a bit like just getting some clay out and putting these together from memory.
There are a few current promising techniques for 3D generation that incorporate ML (NerV etc.) that look promising in the future for replicas of stuff that doesn't exist anymore, e.g. Jabba. however mesh generation still isn't quite there for those.
I think that is just the beta. It only allows one photo so it isn't even hitting full stride yet. I did one of a sculpted horse of the ringwraiths and it is beautiful. They have a second one that does video so I think I can string some photos in sequence and try that one for full wrap around view. For props, it currently needs touchup but for one sixth and gaming objects, it is pretty close to print and paint. I made quite a few more after, bows, shields, helmets, all passable under some paint. The elven helmet even had a head inside, pretty sure it proves they are grabbing existing photos supplied elsewhere.