Ask HN: Best Way to 3D Scan?

Seems like most of the light-based hardware / software is ok but not great for finer details? These sculptures have a lot of vent and keyboard details that are under a millimeter so not sure that path is going to work?

There seem to be a lot of mail away services which I'm totally down to pay for but would really like to get a recommendation on. I have resin copies of all these sculptures so no preciousness issues there.

Thanks in advance! :^)

First problems with 3d scanning are many. One, its just not very accurate. Two, it cannot capture complex geometries such as anything that is occluded. Three, size. You could need a different system for different scales of items to be scanned.

In almost every case the answer is probably photogrammetry. There's an open source project called "OpenScan" which will allow you to build your own turntable/camera system which can take regular photos so you can do a solve.

Before doing this however, I highly suggest you get the best camera you can borrow/beg/steal and go to town with Meshroom. It will cost you a few hours of time/compute power but is otherwise free. There are dozens of free and paid photogrammetry apps whose results vary. I like Meshroom personally.

The reason this technology is underdeveloped is that 3d modeling is cheap and efficient and relatively easy to learn. I was baffled by 3D modeling for most of my life, but once I committed to learning, it wasn't a big deal. The issue is, there are at least half a dozen different types of modeling and tooling, and visual effects are often entwined with all of this. If I may suggest, if you head down this route, use Fusion 360 or Solidworks for mechanical stuff (products, machines, etc.), and Blender or Z-Brush for flowing character things.

Unless your items are artistic in nature and need to be preserved exactly-- you're far better off simply modeling them. As weird as it sounds to say-- your models will be more accurate than the 3D scans. A pair of calipers and a micrometer are way more accurate than current sensors.

Even the most accurate 3D scanning I am aware of (at a prosumer level) will need its mesh edited to make sense. So you're already back to having to learn some form of 3d modeling anyways.

I do believe in 10 years everything will be different-- the big new iPhones have fairly interesting sensors in them but they're only really good for like room level scales. Not very good at small scale things. I imagine as the tech progresses we will see a "sensor-fusion" tech develop for the LIDAR stuff just like we have for the cameras-- IE that it works a multiple focal lengths with the input from those sensors being combined.

Also, if I am out of touch here I'd love to be wrong-- so don't hesitate to blow me up.