TikTok, the short video streaming platform is now offering its users a basic AI text-to-image generator in its newest filter called ‘AI greenscreen’. This filter lets the user input some text and creates abstract images that can be used as the background for their video content. This is also great for content creators.
The company seems to join the trend of AI text-to-image generators after DALL-E 2 got extremely popular and people were eagerly waiting to get through its waitlist. These image generators reached another level of popularity when DALL-E Mini, now rebranded as Craiyon became available to everyone on the internet and people started using it to create hilarious and funny memes. There are several other AI text-to-image generators like Google’s Imagen and Midjourney’s eponymous software in the list too but TikTok seems to be taking it the fun and light way.
The images generated by TikTok’s filter are abstract and produce more vivid color patterns while text-to-image generators like DALL-E 2 are more accurate and are not available to the world for a reason.
The images generated by TikTok’s filter are abstract and produce more vivid color patterns while text-to-image generators like DALL-E 2 are more accurate and are not available to the world for a reason. Imagine if TikTok created something like these models and released it to the world. It would have definitely led to some harmful results. The filter generating such imagery might be intentional so that people don’t misuse the tool and try and keep it fun. Recently, people are following a trend using this filter where they input their name and see the kind of aesthetic the filter generates.
The Verge tried prompting inputs that involved nudity and violence and the results that the filter generated was not photorealistic and rather harmless. One can make out from the color scheme that the generator is getting somewhere but it produces nothing to be too sure about.
One thing this filter suggests is that technology is moving faster than we can imagine. It was only a year ago that DALL-E released its model and now TikTok has also tried its hands on AI. Of course, we are still unaware of the potential of these AI text-to-image generators, we have seen a sneak peek of what people can do if they get access to it. Like Open AI gave access to DALL-E 2 to around one million people recently and one user made it to the headlines by generating images to the prompt “the selfie of the end of the world”.