A method and system are presented for use in real-time insertion of visual effects in a video stream. At least one reference image is used that contains a region of interest, which is to be involved in the video stream. the reference image is processed by defining a set of at least a few reference features therein, and inserting the visual effects into the reference image with a desired orientation of the visual effects with respect to the region of interest in the reference image. Successive video frames are grabbed in a sequence of a few video frames, and at least one of video frames is used as at least one snapshot image. The snapshot image(s) is/are process by selecting therein at least two from the reference features and determining a reference transformation between the snapshot image and the reference image. Independently of this processing of the snapshot image, the snapshot image and other video frames are processed to determine a sequence of transformations between them. These data are utilized to determine an operative transformation between the reference image and either one of the video frames, thereby enabling real-time management of the appearance of the visual effects in the video stream by using the operative transformation for the insertion of the visual effects in other video frames.