Social networks are plagued by curious and strangers videos of celebrities eating and the truth is that it is logical why. Developed by a collaborative team at Huggingface, Modelscope is a “multi-stage text-to-video broadcast model“, which takes text prompts in English, tries to understand what you expect to see and then generates a crazy video.
From Will Smith eating noodles, Scarlett Johansson gobbling spaghetti or Dwayne Johnson destroying the planet’s rocksThe results are worth going viral.
What’s really fun, and the reason there are so many similar videos, is that the AI-powered app is designed to be easy to use and doesn’t require any prior knowledge or experience in machine learning.
ModelScope is capable of generating a variety of video formats, including short videos, animated text, and other visually appealing content—and it’s open source.
Will Smith eating spaghetti or Dwayne Johnson devouring rocks
This tool went public last week and as soon as it appeared on the internet people started making small videos and get the most out of it. Of course, here the great problem of the deepfakes They are set aside because, as you may have already seen, the creations are science fiction —and really fun.
In the case of Will Smith, his creator explains how he did it. “The workflow to create the video was pretty simple: I gave ModelScope the message ‘Will Smith eating spaghetti’ and it rendered it at 24 frames per second (FPS). Chaindrop then used the Flowframes interpolation tool to increase the FPS from 24 to 48 and then slowed it down to half speed, resulting in a smoother video.”
Currently, this technology based on artificial intelligence uses images from stock photo sites and images from Internet and watermarks Shutterstock. However, the developers said that the technology is likely to come to fruition very soon.
Since all these famous people appear eating, the tool has become a viral success and the Internet has been honored with the most curious videos. The great advantage of all this is that there is material for future memes for a long time.