TeDiO: Temporal Diagonal Optimization for Training-Free Coherent Video Diffusion

Introduces TeDiO, a training-free inference-time method that reinforces smooth temporal diagonals in self-attention maps of video diffusion transformers to improve temporal coherence and reduce flickering.