My Post-123D World of 3D Scanning

A 3D-scanner in your pocket was an incredible thing to promise, even for the 21st century. But for years, it was an easy promise to keep, thanks to Autodesk's 123D Catch. I could tell parents, kids, educators, and professionals that it existed, and they'd find it on their own and figure it out.

But in a change of direction, Autodesk discontinued their 123D line of apps and services, leaving a big hole. There must have been hundreds of tutorials on how to do a 3D scan with 123D Catch. And until recently, that included Kacie Hultgren's 3D Scanning with a Camera course on LinkedIn Learning.

The loss of 123D Catch sent us hunting for a replacement. As it turns out, there are a lot of options. For Kacie's update, we chose Autodesk ReCap and Autodesk ReMake. The workflows are very similar to that of 123D Catch: capture a large number of photos while orbiting the object, and use software to construct a 3D model of the object. Kacie introduces the concept, known as photogrammetry, in this video from the course:

3D Scanning with a Camera

The sequence of photos is essentially a very low frame rate video of an orbit around an object. The photogrammetry process determines three-dimensional structure from motion. The artist Rich Oglesby used this to generate 3D scans of mannequin challenge videos, which were relatively high frame rate videos that orbited a scene.

In my search for a 123D Catch replacement, I came across a few other interesting resources.

COLMAP

This is the software that Rich Oglesby used in his project. Unlike 123D Catch, ReCap, or many other tools that do their processing in the cloud, COLMAP runs entirely on your computer (ReMake can also work in a local mode). It requires a lot of computational horsepower, and my 4GB Kepler-era Quadro wasn't up to it. It's somewhat manual, but it gives you tons of control. I intend to spend more time with it to see what I can get it to do, perhaps after I get a computer with a better GPU.

Dr. Peter Falkingham

Dr. Falkingham is a biologist who researches animal motion, and he's really into scanning things (for science!). His blog is full of in-depth posts about different photogrammetry solutions, including many open source solutions. That's where I learned about COLMAP.

Bevel 3D

I backed this Kickstarter some time ago, and conveniently, the Bevel arrived just as I was looking into 123D alternatives. It is a laser scanning attachment for your cell phone, and although it's promising, my first scans were a little strange. I expect it will improve over time. Instead of using photogrammetry, it uses a laser in combination with the cell phone camera to construct a 3D model. It's not really designed to capture 3D objects for fabrication or visualization, but it's pretty neat.

3DFlow Zephyr

This is high on my list of things to try out. They have a free version as well as paid versions, and it performs all its calculations on your GPU, so you don't need to upload images to the cloud to process them.

Skanect

This is what I use for scanning with an ASUS Xtion Pro Live (an older depth camera based on the same hardware as the early Kinect). It uses a different method than photogrammetry to construct models, projecting an infrared pattern on the scene you're scanning and using the reflected dots to determine how far away a feature or object is.

These are some of the things either in my toolbox or likely candidates for my toolbox. There's a lot more out there, though. What are you using to capture reality? Have you found anything that compares to the point-and-shoot ease of 123D Catch?