Also TikTok presented a AI text-to-image, which is a tool that allows you to create an image from a textual description. For example, you could write “Elvis Presley riding a T-Rex” to get an image to go with this scene. And so on.
Text-to-image AI isn’t always very accurate and sometimes the results seem straight out of a horror story, nevertheless they have become hugely popular lately, mainly thanks to the popularity of platforms such as DALL-E by DeepMinda spin-off from Google.
TikTok’s AI is part of the special effects freely available to users and is called ‘AI green screen‘. The images thus created can be used by users as backgrounds for their videos. Exactly, like green screen.
The Verge, who has tested the new TikTok feature, wrote that the results are not always very accurate and that it is undoubtedly a less advanced tool than DALL-E or other similar technologies.
Furthermore, the images produced by the AI have a fairly uniform and abstract style. In short, if other tools even allow you to customize the style of the image (returning to the example above, you can also use “pastel illustration of Elvis Presley riding a T-Rex”), TikTok’s does not offer its users huge freedom of choice and the images all look a bit alike.
According to The Verge’s report, the software is also designed to avoid potentially inappropriate results for young audiences, such as sexual images or violent scenes.
Source: Lega Nerd
