Honestly, how far can we come in just twelve months. This time last year if I told you there was a website you could ask to write an essay on MacBeth after which you’d see it written before your own eyes you’d think I was mad. But today, we see examples of how Google is harnessing computing power and Artificial Intelligence (AI) to generate videos
Again, forget ChatGPT or Google’s Bard for a moment, forget even the work being done at Open AI to generate images from a simple text prompt.
Heck, you can forget the amazing AI Generated images coming out of Photoshop too.
VideoPoet sets a new marker for what’s to come, and it’s video.
Billed as a powerful video model, Google has shared examples of just how this will look, and the kinds of inputs required.
Have you ever seen a “A lion typing on a keyboard” – no, so type that into VideoPoet you get this:
Wondering what “A koala playing piano in the forest, soft lighting” would look like? Just ask VideoPoet:
And these are examples of VideoPoet creating stuff from scratch.
Now imagine it takes an existing video like this standard looking caper:
Now instruct Google’s VideoPoet to edit that video – “Two raccoons on motorbikes. A meteor shower falls behind the raccoons. The meteors impact the earth and explode.”
Boom:
Just like Adobe’s Photoshop is able to expand images, Google’s VideoPoet seems capable of doing that with video too.
Take this video – filmed in a square format, needs to look better for a vertical TikTok or something:
Easy – check this out:
Utterly and overwhelmingly brilliant. Now remember how basic ChatGPT was 12 months ago, and what it can do now.
Brace yourselves, 2024 is going to be wild!!
VideoPoet is an internal project only right now, nothing available to the public to play with, though we’re all keen to have a crack!
Anyway, if you’re not impressed, then here’s “A teddy bear in a leather jacket, baseball cap, and sunglasses playing guitar in front of a waterfall.”
Web: VideoPoet
