If you’re still amazed by the responses from ChatGPT or the images generated by Midjourney, you might be a bit more surprised by the new stage of artificial intelligence: generating videos from simple text. This is what the Modelscope tool offers, which is still in its infancy.
Following the same principle as the other AIs, Modelscope allows you to create short videos from a “prompt”, that is, a written instruction. And the first idea of a Reddit forum user was therefore to make actor Will Smith eat spaghetti. And the result, viewed more than 4 million times on Twitter, is downright terrifying.
Another user took up the same idea, this time with the actress Scarlett Johansson, again laboriously eating spaghetti.
In fact, there seems to be a passion from netizens for celebrities who eat everything from pizza to cakes. Modelscope, and its obvious shortcomings, makes them still look very sketchy or even nightmarish.
Many generated videos display a Shutterstock watermark, suggesting that the tool will draw the basic images from online video and image banks to make them work.
At the moment, Modelscope can be used through the HuggingFace platform but it is completely cluttered. Actually, there are other models that are currently being rolled out, but none are yet mature enough, as was also the case with Midjourney or Stable Diffusion, just a year ago.
OpenAI, the creator of ChatGPT and Dall-E, is also working on similar AI which, when implemented, should show much more convincingly what generative AI is capable of in this area.
Source: BFM TV
