Online video archives provide a large amount of multimedia presentation contents through the Internet. But, it takes a long time to find what they really want to watch from a lot of presentation videos. We have been developing a system to summarize multiple presentation contents that match given keywords. In this paper, we intends to obtain a interesting digest including visually attractive scenes instead of generating a sketch of presentation contents. We propose methods to calculate the importance of scenes by considering color distribution and animation occurrence. Finally, we evaluate the effectiveness of the proposed methods.