Want to see a video of yourself that never existed? No?
Well, tough luck.
The chances of that happening have increased dramatically with Runway's Gen-2 feature 'Image to Video'. I can put a photo of you into its prompt and it will animate it - with some very odd results indeed.
Here's a very stock image of me:
I uploaded it into Gen-2 and the video at the top of this page was the output from simply pressing 'generate'.
You can do it with loads of things. Here's a cute robot teacher that I created in Midjourney (PROMPT: 'Cheerful, helpful robot teacher, cute and endearing, pointing with a bright and engaging demeanor. Design emphasizes education and assistance, ideal for a mascot, 3D rendered with high detail, soft color palette. Set against a pure white background, spotlight effect, studio lighting, friendly vibe --q 2 --ar 1:1 --v 5 --stylize 400')
Upon upload to Gen-2, he became:
Now, you don't need me to tell you that this is not particularly useful. However, this kit is learning all the time and in a space of months I'd be confident that the results will be much better.
So let's revisit in a couple of months and try the same experiment and review the results. Steadily, I'd bet significant amounts on it becoming something quite incredible.
Made with TRUST_AI - see the Charter: https://www.modelprop.co.uk/trust-ai
Comments