AI Could Cut Your Video Streaming Data Use by Nearly a Third

Researchers present PRESLEY, a video streaming system that extends prior conference work called ELVIS. Rather than treating every region of a frame alike, it replaces destructive block removal with adaptive in-place degradation under a removability mask, signals per-block strength in a bit-packed side channel, and restores the degraded regions at the client using generative backbones conditioned on transmitted visual priors rather than unconditioned in-painting. At matched rate against its predecessor, PRESLEY achieves a mean -56.4% BD-rate reduction on delivered background quality across 13 rate ladders spanning multiple codecs and dataset families. Against pristine baselines, it delivers substantial bitrate savings (up to -29.4% BD-rate) and superior background quality on 17 of 23 sequences in the target bit-starved regime, while maintaining foreground fidelity bit-exact. The authors also map the theoretical headroom of this architecture: using an exact leave-one-superblock-out combinatorial oracle as an additive empirical bound, they show existing complexity heuristics already capture 83.3% of bit-cost savings, bounding remaining cost-axis headroom at about 5% of total bitrate. They then identify and model the primary unaddressed axis, post-restoration damage, which disperses widely (4.9 to 8.4 dB), and prove it is predictable before transmission (held-out rho = +0.400).