Movia's 3D systems are based on a computer vision technique called shape from silhouette which we implement in Alias / Wavefront Maya, using a MEL script to automatically produce 3D objects from sequences of images and their corresponding silhouette matte images. Here's how shape from silhouette works:

	Matte images of the subject are extruded to form conical volumes.
	A volume is created for each camera in the array.
	All of the volumes are aligned exactly as the cameras were alligned.
	The Boolean intersection of the volumes is a 3D object which is roughly the same shape as the subject.		This 3D object was created using only 8 cameras. Larger numbers of cameras result in more object detail.

Additional techniques are used to increase object detail. For many applications a rough 3D object provides a sufficient base geometry for further image-based processing. For photorealistic rendering objects are simplified and textured with photographic detail from the subject - resulting in low-bandwidth photorealistic renderings without detailed underlying geometry. return to post production overview