TikTok’s new special effects introduce AI model based on Wenshengtu
According to foreign media reports, TikTok’s new special effects introduce a Wenshengtu AI model, called “AI Green Screen”, that is, when the user enters a piece of text, AI will automatically generate a picture that matches the situation. In the past, artificial intelligence company OpenAI, Silicon Valley giant Google, and graphics processor giant NVIDIA have also published similar applications. With the promotion of TikTok, which has a user base of 1 billion, this technology is entering the field of vision of more young people.
Photo credit / Solen Feyissa
This special effect was first discovered by the foreign media “The Verge”, and now you can find the special effect by searching for “AI Greenscreen” on TikTok, but not all models of mobile phones are applicable, such as the author’s Galaxy S10, which cannot be enabled. Nonetheless, the average user can still browse the results of other users.
Screenshot/TikTok
When first started, TikTok will prompt users to try to enter simple words, such as “snorkeling in the purple ocean” and “erupting rainbow volcano”. However, compared with the more mature technology and the OpenAI model DALL-E, TikTok’s special effects do not follow a realistic route, but are headed by abstract-style artistic pictures. Even if you enter specific words (such as country names), the results will only be indescribable. abstract painting.
“The Verge” reported that TikTok’s AI model with Wenshengtu may have certain limitations, such as nudity, violence and other content that violates community rules, and the model will tend to generate abstract images to avoid inappropriate words.
Although TikTok’s special effects are not as good as other models, and they can’t help artists or professionals in practice, AI models such as DALL-E are not yet fully open. With the promotion of TikTok, they can enhance the videos of millions of creators, while Bring fun to more people.