Video · Text → video · Image → video
Wan 2.7 Video
720p and 1080p video from text, frames, or multiple references
About the model
Wan 2.7 is a line of video models from Alibaba’s Tongyi Lab, each covering its own step of the video workflow. Compared with the previous version, it adds first- and last-frame control and generation from several references at once, along with better overall quality and temporal consistency.
Our service offers three modes:
- Text to video — a 2–15 second clip in 720p or 1080p from a description alone, with a choice of aspect ratio.
- Image to video — animate a single frame or a transition between a first and last frame; the model fills in the motion and keeps the subject’s look.
- Reference to video — up to 5 images the model uses to keep characters and objects looking the same in a new scene.
Strengths
- Three modes — from text, from an image, and from references
- First- and last-frame control — the model fills in the motion between them
- Up to 5 references for consistent character appearance
- 720p or 1080p output
- Clips up to 15 seconds (up to 10 seconds in reference mode)
- Intelligent prompt expansion on by default
Best for
- Animating finished illustrations and product photos
- Transitions between two set frames
- Clips featuring the same character across scenes
- Quick social media videos from a text description
Limitations
- Reference mode — up to 10 seconds and up to 5 references
- Image mode doesn't let you choose the aspect ratio
- Minimum resolution is 720p
Prompting tips
- In reference mode, name references by order — "image 1", "image 2"
- In image mode, describe the motion rather than what is already in the frame
- The model expands short prompts on its own; detailed prompts are followed more precisely