Google announced the Omni 1.2 video generation engine so I tested it with the help of my buddy “Da Frog With A Pink Wig”. I tried one-shot text-to-video prompting with four different visual styles and I was not satisfied with any of them as they each had niggling errors that broke the continuity or coherence. For example, the frog’s hand-pounding actions didn’t look right, the eyeballs weren’t consistent, or the weird addition of text in the headline in the last clip.
Of course, I could improve the outcome by uploading still images to do frame-by-frame direction of the AI, but I’m testing this from a layman’s point of view. Like I tell everyone in my AI workshops: “Don’t waste too much time making AI videos, check back six months later to see how the tech progresses.” I’ll see you, Omni, in 2027.
(Videos generated in Google AI Studio and compiled in Canva.)