Hailuo has announced the launch of a new Subject Reference feature, which enables the creation of videos with incredible character identity stability. Now, to generate a video in which a character maintains their identity throughout all frames, only a single reference image is needed.
🎥 Introducing Hailuo's Subject Reference: Revolutionizing Character Consistency in Video Creation 🔥
— Hailuo AI (MiniMax) (@Hailuo_AI) January 9, 2025
We’re excited to present Hailuo's S2V-01 model, a groundbreaking innovation in AI video generation that tackles one of the industry’s biggest challenges: maintaining consistent,… pic.twitter.com/uoZO687xdJ
The feature works in three steps: the user uploads a single image of the character, creates a prompt to set the direction of the video, and then starts the creation process. This allows for videos with a high level of accuracy and consistency of the character’s appearance, even when the camera angle or the character’s movements change.
The uniqueness of the new technology is that it is the first to allow for character consistency in dynamic videos. Each frame reproduces the original character image with maximum accuracy. Significant flexibility has been achieved in changing not only facial features but also poses, facial expressions, lighting, and other details via text prompts.
Although the new model improves character consistency, it may sometimes follow prompts less accurately than T2V or I2V models, especially when the environment changes. However, this is just the beginning — the developers are working on improvements, including multi-reference images for multiple characters and more complex scenes.
