AI Paper Atlas

Research map / Video / Video generation

DLFR-Gen: Diffusion-based Video Generation with Dynamic Latent Frame Rate

Zhihang Yuan, Rui Xie, Yuzhang Shang, Hanling Zhang, Siyuan Wang, Shengen Yan, Guohao Dai, Yu Wang

ICCV 2025 · Video › Video generation · approach: Efficient video diffusion models · 1 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
apply DLFR-Gen dynamic latent frame rate to HunyuanVideoHunyuanVideo · VBench prompt subsetLatency (s): 714.90 -> 245.12Improved
apply DLFR-Gen with strictest threshold theta=0.9, k=15HunyuanVideo · VBench prompt subsetSpeedup: 1.24Improved
lower similarity threshold theta from 0.9 to 0.5 (k=5) (ablation)DLFR-Gen · VBench prompt subsetSpeedup: 1.31 -> 2.92Improved
lower similarity threshold theta from 0.9 to 0.5 (k=5) (ablation)DLFR-Gen · VBench prompt subsetSSIM: 0.631 -> 0.513Hurt
start compression later (k=5 to k=15, theta=0.5) (ablation)DLFR-Gen · VBench prompt subsetSpeedup: 2.92 -> 1.96Hurt
start compression later (k=5 to k=15, theta=0.5) (ablation)DLFR-Gen · VBench prompt subsetVQA aesthetic score: 97.85 -> 96.57Hurt
start compression later (k=5 to k=15, theta=0.9) (ablation)DLFR-Gen · VBench prompt subsetSSIM: 0.631 -> 0.723Improved
DLFR-Gen versus original on temporal flicker (k=5, theta=0.5)HunyuanVideo · VBench prompt subsetTemporal flicker: 0.9793 -> 0.9919Improved
DLFR-Gen versus original on VQA (k=5, theta=0.5)HunyuanVideo · VBench prompt subsetVQA aesthetic score: 98.63 -> 97.85Hurt
add the DLFR-Gen compression module (versus operating directly in latent space) (ablation)DLFR-Gen · qualitative cat video (Fig. 4)visual qualityImproved
add one-step denoise and renoise around the compression module (ablation)DLFR-Gen · qualitative cat video (Fig. 4)visual qualityImproved
start compression at earlier denoising timesteps (ablation)DLFR-Gen · qualitative (Fig. 4)visual qualityHurt

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