AI Paper Atlas

Research map / Video / Video generation

PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference

Xiaofeng Mao, Shaohao Rui, Bo Zheng, Kaining Ying, Chuanhao Li, Mingmin Chi, Kaipeng Zhang

ECCV 2026 · Video › Video generation · approach: Autoregressive long video generation · 18 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 sink tokens (N_sink 0 -> 8) (ablation)PackForcing · 60 sampled promptsoverall CLIP score: 31.24 -> 35.09Improved
add sink tokens (N_sink 0 -> 8) (ablation)PackForcing · 60 sampled promptsSubject Consistency: 74.72 -> 93.11Improved
increase sink size from 8 to 16 (ablation)PackForcing · 60 sampled promptsDynamic Degree / CLIP: Dynamic Degree 49.84 -> 35.16; CLIP 35.09 -> 35.39Mixed
use both HR and LR compression branches vs HR only (ablation)PackForcing · VBench-Long 60sCLIP: 32.97 -> 33.54Improved
use both HR and LR compression branches vs LR only (ablation)PackForcing · VBench-Long 60sOverall Consistency: 25.18 -> 26.07Improved
dynamic context selection instead of FIFO eviction (ablation)PackForcing · VBench-Long 60sSubject Consistency: 87.82 -> 88.62Improved
dynamic context selection instead of random selection (ablation)PackForcing · VBench-Long 60sSubject Consistency: 86.31 -> 88.62Improved
add incremental temporal RoPE adjustment (ablation)PackForcing · VBench-Long 60sCLIP gap early vs late: 2.53 -> 0.95Improved
three-partition compressed cache vs Self-Forcing for long generationPackForcing vs Self-Forcing · VBench-Long 120sSubject Consistency: 74.40 -> 92.84Improved
three-partition compressed cache vs LongLivePackForcing vs LongLive · VBench-Long 60sSubject Consistency: 92.00 -> 90.49Hurt
three-partition compressed cache vs LongLivePackForcing vs LongLive · VBench-Long 60sDynamic Degree: 44.53 -> 56.25Improved
three-partition compressed cache vs Deep ForcingPackForcing vs Deep Forcing · VBench-Long 120sBackground Consistency: 92.55 -> 91.88Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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