ByteDance Launches Limited Test of Seedance 2.0 Video Generation Model
ByteDance, the parent company of popular AI assistant app Doubao, has begun a limited rollout of its latest video generation model, Seedance 2.0, for testing purposes. The move signals continued investment in artificial intelligence and content creation tools by the Chinese tech giant, which operates globally.
According to reports on February 11, users within the test group accessing the “AI Creation” feature in the Doubao app, and selecting “Video Generation,” now have the option to utilize the Seedance 2.0 model.
Currently, Doubao app users can leverage Seedance 2.0 to create videos lasting either 5 or 10 seconds. In comparison, the company’s all-in-one AI creation platform, “Jimeng,” allows for video generation ranging from 4 to 15 seconds.
The Doubao app interface indicates that generating a 10-second video using Seedance 2.0 consumes two video generation credits, with users having a daily allowance of eight credits.
The launch follows a soft launch of Seedance 2.0 on February 7, boasting support for four input modalities – images, videos, audio, and text – with its “reference capability” highlighted as a key strength.
Feng Ji, founder of Game Science and producer of “Black Myth: Wukong,” recently lauded Seedance 2.0 as “the strongest on the surface, bar none” following an in-depth evaluation on February 9, as reported on his Weibo account.
The positive reception extends internationally, with overseas users reportedly seeking access to the model and sharing registration tutorials online, as Seedance 2.0 is not yet available in all regions.
Early testing indicates that Seedance 2.0 demonstrates improved comprehension of complex prompts and delivers more natural audio-visual synchronization and overall visual quality compared to ByteDance’s previous generation video model.
ByteDance’s official technical documentation highlights the model’s ability to accurately follow lengthy prompts, recreating complex scenarios such as difficult musical performances and surreal imagery.
