RSS
@NYC Toy Fair 2009 - One of the more interesting companies was 3D3 solutions, they use off the shelf web cams and presentation projectors to make a 3D model, you can see me getting 3D scanned above - good stuff and not as expensive as many of the other commercial packages.
FlexScan3D is an innovative non-contact 3D scanning software system that takes 3D measurements and creates digital 3D models directly from physical objects. The software is ideal for measuring complex shapes quickly. FlexScan3D is the world's only fully customizable 3D scanning system that scans objects using off-the-shelf hardware components. It is a cost effective solution that is versatile in many applications. How FlexScan3D scanning works: Using one or two digital cameras and a white light (presentation) projector FlexScan3D will scan a wide range of objects, faces, and body parts quickly and accurately. The projector or laser puts reference patterns on to the scan target to aid accurate digitalization. The scene is captured using a camera or video camera. FlexScan3D's 3D triangulation engine takes these images and creates millions of measurements of the scene. The process to acquire the data needed to create a 3D model is then completed in seconds. The 3D scan data is used in industrial design, reverse engineering, visual effects, and biomedical industries. Automated 3D capture drastically reduces the time and cost in capturing complex physical measurements.
Please get your face model from them and post it online Phil!
Reply to this comment
This is called 'space coding' or something. Basically each horizontal line (I think usually they do it vertically) is given a number, this is converted to binary, and then it just projects the "first bit image", then "second bit image" and so on. E.g. for eight lines you have:
0 000
1 001
2 010
3 011
4 100
5 101
6 110
7 111
So the first image is made of the first bits: 00001111, then 00110011 and so on.
Now when you look at a single output pixel in the video you can easily work out which line was shining on it. E.g. if it is bright-dark-bright then it must be in line 5. Then simply triangulate the line that the point must be in from the camera, and the plane it must be in from the projector to get its 3D position.
I'm surprised there is no open source software for this, it's probably a couple of hours of work. Anyone know of any?
Reply to this comment
Actually, thinking about it I wonder why they didn't use gray codes... They'd probably work better. Hmm I might try this.
Reply to this comment
I wonder if this would work with infrared light or an infrared scanner.
I'm imagining a "Mission Impossible" scenario, where you 'scan' someone without their knowledge, make a 3d model of them, and create a mask to impersonate them.
OK, the mask part is kinda farfetched. But I still wonder about the infrared scanning. It might be less instrusive than having the white light shining in your eyes.
Reply to this comment