Ray 2 is a video generation model from Luma AI, the predecessor to Ray 3 in Luma's generative video model lineage. Released as part of Luma's progression of video generation capabilities, Ray 2 represented a significant advance over earlier Dream Machine outputs in terms of motion quality, visual coherence, and the range of cinematic styles the model could produce.

Ray 2 improved on its predecessors in motion smoothness, temporal consistency, and the ability to follow complex prompt descriptions involving camera movement and scene dynamics. The model supported both text-to-video and image-to-video generation, allowing creators to anchor generations to reference images while specifying the motion and scene development through text. As with other iterative model releases, Ray 2 established benchmarks in generation quality and capability that informed the subsequent development of Ray 3. The model demonstrated Luma AI's focus on cinematic quality and creative expressiveness as differentiating characteristics of its video generation approach, with particular attention to lighting fidelity, depth, and the realistic behavior of dynamic elements such as water, cloth, and environmental effects.

For creators tracking the development of AI video generation tools, Ray 2 represents an important point in Luma AI's model progression, illustrating the rate of capability improvement across successive generations of the same model family and providing context for the advances that subsequent releases in the lineage introduced.