This was generated using stable diffusion with the prompt starting with "South Dakota". It is one in a series of shorts exploring the underlying dataset that it was trained upon. The jigsaw puzzle pieces are confusing...
This was generated with stable diffusion using "Kid Cudi" as the prompt to attempt a Being John Malkovich like vibe. There is no motion model here so there are temporal artifacts frame-to-frame that are amplified by the low-resolution video.
Thanks for the comment. Yeah the song is great. Do you like the trailer better: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-BOLso4HVKIw.html. it is a work in progress.