Selecting The generation of movie preview has been studied in 2D field as a technology, which allows us to skim summarized videos in a short time. However, directly implementing on existing 2d preview method in S3D contents lead to many problems because S3D human factors that can cause visual fatigue are not considered in 2D preview methods. This paper introduces a novel method that selects highly depth-perceived scenes and simultaneously minimizes visual fatigues with S3D human factors, which are not considered so far to create effective stereoscopic 3D movie preview. First, to rank the scenes that have minimum visual fatigue with high depth-perception, we scored each scene based on three S3D human factors: luminance, blur, and motion, as well as depth range. Second, we implemented our simple and robust approximation algorithm to accomplish the depth stabilization and the arrangement of scenes at once. We have applied our novel method to movies of different characteristics for evaluation. Experiments show that our technique dramatically preserves the good continuity of depth while extracting effective S3D scenes.