AI CINEMA

The course offers a structured survey of the current state of generative moving-image production, examined at the level of technique rather than novelty. Sessions proceed through the principal classes of system now in use — text-conditioned, image-conditioned, and video-to-video generation — and through the control mechanisms that have developed around them: reference and keyframe conditioning, camera and motion specification, identity persistence across shots, performance and lip-sync transfer, temporal upscaling and interpolation, generative sound and score, and the integration of all of these into a conventional editorial and grading pipeline. Throughout, emphasis falls on the failure modes characteristic of each method and on the strategies by which those failures are concealed or exploited. The course proceeds from the position that technique in this field consists not in fluency with any particular instrument, but in judgment about what each can be made to do.

Attendance during the whole seminar.