FloED uses optical flow as extra motion guidance in a diffusion video inpainting model, with flow-warped latent interpolation and attention caching to cut inference cost while improving temporal consistency.
Two birds, one stone: A unified framework for joint learning of image and video style transfers
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Coherent Video Inpainting Using Optical Flow-Guided Efficient Diffusion
FloED uses optical flow as extra motion guidance in a diffusion video inpainting model, with flow-warped latent interpolation and attention caching to cut inference cost while improving temporal consistency.