The fact that we’re dealing with warped materials, parallax issues, and fisheye lens corrections, it seems like a tall order to get anything approaching real-time precise alignment with the cameras… there are just so many variables. I don’t doubt that in an ideal situation it could be overcome, I just wonder two main things: Can the camera hardware precision we have now achieve that and is it worth the development cycles to chase it?
Add to that their server-based model that does the heavy lifting and now you’re on the hook for a detailed material scan (to account for warp) and upload to the server… all of which puts us firmly in the “I should have jigged” time category. I haven’t timed it, but my best recollection is that a proper jig workflow for a single item adds maybe 2-3 minutes to the job time (setup/breakdown of jig plus cut times in the 8-10 second range) and achieves conservatively 0.01" accuracy. I should time it next time I run a job, just to see.
What level of precision do you think is reasonable? I typically get to about a mm or two with my cameras ( I know there are people with much worse alignment), if I need more than that, it’s jig city. Maybe your tune will change once you have your machine, but I suspect its not as bad as you’re imagining it to be. I’m actually realizing now that I’m probably telling you a lot of stuff you don’t really need to know about yet since I don’t think you even have your GF yet… maybe we can pick this back up once you have your machine and can speak from experience?