A couple weeks back I was scrolling through Instagram and stumbled on a post from Raullinaitis that genuinely made my eyeballs almost pop out of my skull. The “videos” he posted weren’t even videos—they were product photos that got dragged and dropped into an AI tool and somehow turned into these surreal, moving shots. So in this video I jumped into Runway’s Gen-2 to see what it can actually do, and I took you along for my first real go-around. I tested both text-to-video and image-to-video. With text prompts, it was equal parts fascinating and completely bizarre—cocktail glasses turning into impossible shapes, an astronaut with three legs, and pizza that somehow looked like a cookie. Then I tried feeding it my own product shots and screenshots from my old videos. You can see flashes of potential—stuff that would be basically impossible to do in-camera—but the AI still mangles details like product text, invents random elements (it literally added a call-to-action I didn’t ask for), and sometimes gets… terrifying (especially when it tried animating my face). My takeaway: Gen-2 is interesting and fun right now, and I’d love to fast-forward 10 years to see it perfected—because there’s clearly something here.