Due to the limited budget and high maintenance cost, it is difficult to deploy a large amount of roadside units (such as 802.11 base stations) to cover the complicated road areas. This paper addresses an important topic on frame sharing since vehicles often need to download large size data such as video clips regarding local news. It is important to fairly use V2V communications at places where an infrastructure is not immediately available or cannot be directly used. We aim to improve local video downloading performance by integrating V2V and V2I communications. The proposed approach, ZFEP/SD, is able to exploit user requests locality and to initiate video frames sharing among neighboring vehicles. We also consider the case of selfish users, e.g., those not willing to share frames with others. Numerical results show that our ZFEP performs better than a pure infrastructure-based download, and that ZFEP/SD ensures that cooperative nodes always get better service than every type of selfish nodes.