AI Paper Atlas

Research map / Video / Video generation

Lightning Fast Caching-based Parallel Denoising Prediction for Accelerating Talking Head Generation

Jianzhi Long, Wenhao Sun, Rong-Cheng Tu, Dacheng Tao

AAAI 2026 · Video › Video generation · approach: Efficient video diffusion models · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add input latents estimation to parallel denoising prediction (ablation)LightningCP on EchoMimic · HDTF subsetFVD: 460.21 -> 225.16Improved
add input latents estimation to parallel denoising prediction (ablation)LightningCP on Hallo · HDTF subsetFVD: 208.46 -> 200.33Improved
add input latents estimation (ablation)LightningCP on EchoMimic · HDTF subsetE-FID: 0.226 -> 0.233No clear effect
add Decoupled Foreground Attention on top of LCP (ablation)LightningCP on Hallo · Hallo FLOPsFLOPs (x10^12/clip): 745.72 -> 679.56Improved
add Decoupled Foreground Attention on top of LCP (ablation)LightningCP on Hallo · Hallo speedupspeedup: 2.94 -> 3.08Improved
add reference feature removal (ablation)LCP + DFA on Hallo · MEADSync: 6.19 -> 6.43Improved
add reference feature removal (ablation)LCP + DFA on Hallo · Hallo speedupspeedup: 3.08 -> 3.15Improved
add DFA on top of LCP (ablation)LightningCP on Hallo · HDTFFID: 4.99 -> 5.10No clear effect
use LightningCP + DFA + rm instead of FasterDiffusiontalking head diffusion models · HDTFFVD: 140.56 -> 121.80 (Hallo)Improved
use LightningCP + DFA + rm instead of DeepCacheHallo · Hallo speedupspeedup: 1.84 -> 3.15Improved
fewer diffusion steps limits non-key timestepsMEMO and EchoMimic Acc. · Speedupspeedup: 3.15 (Hallo) vs 2.33 (MEMO) vs 1.45 (EchoMimic Acc.)Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy