A series of examples playing with video generative AI. The following samples three different tools (Runway, DeepAI, and invideo) attempting to animate a '40s photograph from a small town in West Virginia, the West Virginia state coat of arms, and the state flag, respectively.
A first iteration of merely describing the scene wasn't nearly interesting enough, so I prompted again in a second iteration by describing a more panicky scenario. With only slightly more movement, this is nonetheless better:
After a failed attempt to get Runway to animate the coat of arms, I tried other tools. I described the scene in hopes that, even though it was cartoonish, something might get picked up and animated. This was the only example across any tool that came close.
Hoping that some model might hold some information within that it can access despite not being provided a source, much like I'd taken advantage of with ChatGPT, I sought out a text-to-video generative AI. I did at least discover how all those spammy videos on social media are generated so quickly.