AI art is taking the creative world by storm, whether you like it or not.
Meta is now furthering its reach in the sphere of artificial intelligence with , a DALL-E-like tool that creates artistic videos from scratch from text prompts.
With Make-A-Video, all you would need to do is put in a few simple words or a short description of your vision and wait for it to be churned out.
“A teddy bear painting a portrait.” Image via Meta
To showcase the ability of its new model, Meta created a few test videos and GIFs of various scenarios, including a teddy bear painting a picture with a real hand attached to it, an astronaut floating in space, unicorns racing down a beach, and a ballerina on top of a skyscraper.
“A golden retriever eating ice cream on a beautiful tropical beach at sunset, high resolution.” Image via Meta
The scenes have a surrealist quality that makes you feel like you are watching a dream, as the movements and colors are slightly off-kilter.
“A blue unicorn flying over a mystical land.” Image via Meta
The videos were developed with deep learning from synthesizers such as GPT-3 and LAION-5B. However, according to Meta, the AI machine will only be able to produce short, albeit high-quality, scenes.
While turning sentences into moving images is pretty cool, the new generator will also be able to animate still images. Just upload a picture of your choice and watch it spring to life. One example shown is a painting of a ship caught in a storm out at sea. The system then animated the waves, and soon, the boat bounced through the ocean.
Image via Meta
But before you rush off to try your hand at creating your scenes, Meta has not officially released Make-A-Video yet. As of right now, it has just been announced that the model is still in the works.
However, there is a dedicated website for the Make-A-Video tool, where you can see other instances of its prowess and read a research paper on the whole thing.
There’s no denying how far digitally-generated art has come from DALL-E 2 to the first movie made entirely out of AI and now a system that might potentially allow everybody to create moving dreamscape scenes with the push of a button.
“A confused grizzly bear in calculus class.” Image via Meta
“A ballerina performs a beautiful and difficult dance on the roof of a very tall skyscraper; the city is lit up and glowing behind her.” Image via Meta
[via PetaPixel and Meta, images via Meta]
http://www.designtaxi.com/news/420576/Meta-Flaunts-Freakish-Make-A-Video-AI-Tool-That-Creates-Moving-Art-From-Text/
Leave a Reply