Stereoscopic 3D view synthesis from unsynchronized multi-view video

We present an alternative approach to flexible stereoscopic 3D video content creation. To accomplish a natural image look without the need for expensive hardware or time consuming manual scene modeling, we employ an image-based free-viewpoint system to synthesize the stereoscopic views. By recording the sequence in a sparse multi-view setup, we are able to maintain control over camera position and timing as well as the parameters relevant for stereoscopic content. In particular, we are able to use the system to match camera path and timing of time lapsed background footage and a live-action foreground video.