Systems based on artificial intelligence that can generate images from a text description are becoming increasingly popular. Evidence of this is the appearance of such a tool on the popular short video service. tick tock.
Not so long ago, a new effect called “AI Green Screen” appeared on the platform. Interaction with this tool requires that the user enters a text description of the desired image in a special window, after which the algorithm based on it creates a full-fledged image. In the future, TikTok users can use the images generated in this way as the background for the video, which makes the new effect interesting for content creators.
Note that the capabilities of TikTok’s AI algorithm are quite modest compared to more advanced systems like Google’s Imagen or OpenAI’s DALL-E 2. The tool available to users of the platform is capable of generating only abstract and obscure images. The algorithm is therefore not suitable for creating photorealistic images or complex illustrations.
Limiting the capabilities of the AI green screen tool may be intentional as more processing power is required to create photorealistic images. In addition, TikTok has over a billion users, so it’s hard to imagine what each of them might think about drawing. Probably, the developers first made sure that users would not be able to create images that in one way or another violate the rules of the service.