Moonvalley

Generates cinematic videos from text or images with precise motion controls. a research lab focused on pushing the frontiers of creativity through deep learning. Moonvalley is a platform using the Marey model to generate cinematic videos from text or images, focused on professional production with licensed training data. Marey trains exclusively on fully licensed, high-resolution footage to avoid copyright issues in generated content. It accepts text prompts, single images, reference videos, and drawn trajectories for video generation. Camera Control creates 3D-aware pans, zooms, and dollys from a static image input. Clips generate up to 30 seconds, with extensions available for longer sequences. Basic access starts with limited credits on Standard plans, allowing tests without full commitment.