ByteDance, the Chinese company behind TikTok, is stepping up its push into the Korean market with a new artificial intelligence (AI) video generation model.
According to the information technology (IT) industry on the 6th, ByteDance's new AI video model "Seedance 2.5" began official service in Korea this month. Domestic users can now access it through "CapCut," a video editing platform operated by ByteDance.
Seedance 2.5 is the successor to the previous version, Seedance 2.0. Shortly after its release in February, Seedance 2.0 took first place in the video generation benchmark of Artificial Analysis, an AI model performance analysis firm, causing a significant stir in the global AI market.
The new Seedance 2.5 is notable for its ability to generate videos of up to 30 seconds, double the length of the previous model's 15 seconds. Users can produce a one-take video of up to 30 seconds from just a single piece of text or image, and can also perform detailed edits such as background replacement, object removal, and style changes on existing frames.
In addition, using "video extension mode," a 30-second video can be lengthened to up to 60 seconds, while "ultra-long mode" allows the creation of videos up to 180 seconds (3 minutes) in length. In particular, the model has significantly improved chronic problems that frequently occurred in existing AI video production, such as glossy human skin, clumped animal fur, and tangled complex movements.
The most noticeable change from the previous model is the addition of a "timestamp text control" feature. For example, when given a natural-language command such as "have the person turn around at 5 seconds and switch the screen at 7 seconds," the AI precisely recognizes the timing and performs the requested action and staging at that point.
ByteDance included Korea among the initial launch countries for Seedance 2.5. Following Seedance 2.0, which made its domestic debut in March, the next-generation version has been introduced to the Korean market just five months later.
