In this work, a vehicle-to-vehicle (V2V) visible light communications (VLC) model for two practical scenarios, is proposed. In scenario 1, the random lateral shift of vehicles and the deterministic longitudinal separation between two communicating vehicles are considered, whereas in scenario 2, longitudinal separation between two vehicles is considered to be random, and lateral shift of vehicles is considered to be deterministic. To this end, we emphasize comprehensive modeling of the practical characteristics of the considered V2V-VLC system, such as random path loss due to the random mobility of the vehicle, random lateral shift and random longitudinal separation of the vehicle. Moreover, we analyze the performance of the proposed V2V-VLC model in terms of different metrics under the consideration of a novel channel model. Considering our findings, it is observed that the random lateral shift of the vehicle and the random longitudinal separation between two vehicles have a significant impact on the V2V-VLC system performance. Further, at a distance of 40 m, for example, the path loss penalties for moderate and dense fog weather scenarios are 2 and 3 dB, respectively, compared with the clear weather. Furthermore, the combined impact of path loss and atmospheric turbulence affects the V2V-VLC performance significantly.