This event explores how to bring more creative control into AI video using Martini and Gaussian Splats. For three years, AI video has had a strange problem: you don't really direct the camera. You ask for a camera move and hope for the best. Martini aims to change this by allowing users to reconstruct a single still image into a navigable 3D space, move a virtual camera through it, choose lenses, set keyframes, and block movements before generating video. The session will include a filmmaker's introduction to Gaussian Splats and Martini's approach compared to prompt-based camera controls. Participants will work in teams to choreograph different camera moves (creep-in, reveal, orbit, push-through, wildcard) using the same still frame, then generate and compare the shots. The goal is to determine how much control blocking a shot provides versus prompting, and where this technology fits into a filmmaker's workflow for previs, pitching, client work, or finished shots. The event emphasizes hands-on making over theoretical discussion, teamwork, sharing compute power, and collaborative learning.