Motion-I2V: Consistent and Controllable Image-to-Video Generation with Explicit Motion Modeling.
Xiaoyu ShiZhaoyang HuangFu-Yun WangWeikang BianDasong LiYi ZhangManyuan ZhangKa Chun CheungSimon SeeHongwei QinJifeng DaiHongsheng LiPublished in: SIGGRAPH (Conference Paper Track) (2024)
Keyphrases
- object motion
- static images
- space time
- image frames
- camera movement
- temporal continuity
- image motion
- visual data
- visual cues
- successive frames
- key frames
- image sequences
- input video
- layered representation
- reference frame
- visual perception
- feature points
- temporal filtering
- video sequences
- image features
- normalized correlation
- image retrieval
- motion analysis
- multiscale
- image data
- single frame
- dynamic scenes
- single image
- motion field
- camera motion
- spatial and temporal
- position and orientation
- motion capture
- rotational motion
- human motion
- motion segmentation
- moving objects
- optical flow
- displacement estimation
- motion estimation
- input image
- temporal domain
- motion model
- video signals
- video scene
- motion features
- adjacent frames
- image classification
- video frames
- ground plane
- target object
- video data
- video streams
- dynamic textures
- video content