arXiv cs.CV
· Papers
Chorus II: Cross-Request Sparsity Reuse for Efficient Image-to-Video Generation
arXiv:2606.25040v1 Announce Type: new Abstract: Serving diffusion models for image-to-video generation is computationally expensive, posing significant challenges for large-scale deployment. Real I2V workloads often contain similar requests, such as repeated effect templates, related subjects, and recurring shot layout