What Is Pose Transfer?
In high-end production, performance is everything. Every pose, gesture, and expression shapes the story. Pose Transfer lets you define how a character moves and reacts by adjusting their body position and facial expression frame by frame.
Use reference images or performance inputs to guide posture, emotion, and timing, and apply those details to any human (animals coming soon!).
How It Works
Pose Transfer lets you define how a character moves and reacts by providing a reference video that will then be applied to an image of the character. Reference videos can guide posture, emotion, and timing, and apply those details to an image of a human or animated human to create a video output.
For example, you can apply a video of a person dancing to an image of a monkey in order to make the monkey come alive and perform the exact dance in the outputted video.
Here's how its done:
Select Pose Transfer from the main dashboard
Upload your Input Video (this is the video that will inform the motion of the generation) and select Open.
Text-to-video OR image-to-video: You can either write a prompt to describe a brand new video that will have the motion of the uploaded video applied (text 2 video) OR you can upload a keyframe with a scene you already like.
If you're using a keyframe, upload the image to apply the pose to (this is the character that you want to be shown in the actual generation). Write a prompt to describe what you want to happen in the video, including a description of the keyframe and the motion being added to it.
IMPORTANT: The more consistent the composition of the Input Video and Keyframe is, the better the results.
Hit generate and wait for your video to generate for review. If you aren’t happy with the results, your conditioning video and keyframe image may need to match more closely.