In computer vision 2-dimensional videos have been widely used. However, the distance and depth of the objects cannot be determined in 2D due to lack of information provided in the video. Though pseudo 3D model developed using stereoscopic video combining multi-frames from different angles gives a sense of 3D, it does not provide enough information to accurately determine the depth of the object thus making it difficult to reconstruct the object in 3D. The Kinect sensor is capable of overcoming all these problems. However, the loss of information due to the presence of noise and flickering effect. In this paper, we present a pixel filtering method to overcome the flickering effects in Kinect frames. The experiments conducted using full user pose tracking in 3D video format show that the proposed methods make the 3D rendering process smooth and also fast enough to be used in real-time applications