In the last couple of years, we’ve seen some really impressive DIY 3D scanning rigs built using the Microsoft Kinect, like a depth-sensing DSLR rig and even a completely handheld 3D scanner. But who says you need a Kinect to power a 3D scanner?
One DIYer named Giancarlo Todone built his own simplified and super affordable $50 system called the DephtInition using a single hacked Canon Powershot and some custom firmware DephtInition is ingenious in its simplicity since it only needs one camera to capture multiple exposures at different focal lengths. The process, known as focus stacking, takes this collection of photos and looks for differences in luminance. Focus stacking isn't a new technique—photographers use it to extend the depth of field of their digital images–but Giancarlo used it in a new and novel way.By analyzing the brightness of the pixels in relation to each other, an algorithm can discern areas of focus and assign “focus ranks” to parts of the image, which allows it to create a depth map. Once DephtInition does that, Giancarlo can then combine his depth map with the RGB images from the camera to create the final 3D imaging point clouds in MeshLab.
The best part about the focus stacking method is it doesn't alter any of the original pixels, so the depth map retains its original color. If you’ve ever seen a Kinect depth-sensing map, it looks more like a grayscale bitmap of that outlines your features.
It’s a super simple and inexpensive method of creating color-accurate 3D images from a stack of multi-focused 2D images. But the method is not without its flaws. For one thing, DephtInition is only useful for capturing subjects that are perfectly still. Meanwhile, the subject has to be illuminated just right for the computer to accurately pick up the depth data.