AI Paper Atlas

Research map / Video / Video generation

Cycle-World: Mitigating Error Accumulation in Long-term Video World Models via Reverse-Prediction Cycle Consistency

Zihan Su, Teng Hu, Jiangning Zhang, Ruiyan Wang, Ran Yi, Lizhuang Ma, Dacheng Tao

ECCV 2026 · Video › Video generation · approach: Autoregressive long video generation · 2 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 cycle-consistent learning (CCL) to baseline (ablation)LongLive · Physical consistency 5sPC: 61.94 -> 68.70Improved
add cycle-guided inference (CGI) to baseline (ablation)LongLive · VBench 5sTotal score: 83.72 -> 84.51Improved
add CGI only (without CCL) (ablation)LongLive · Physical consistency 5sPC: 61.94 -> 65.67Mixed
add CGI on top of CCL (full Cycle-World) (ablation)+ CCL · Physical consistency 5sPACE: 79.5 -> 81.91Improved
add CCL to baseline for 60s generation (ablation)LongLive · Physical consistency 60sPC: 59.45 -> 70.40Improved
add CCL to baseline for 60s generation (ablation)LongLive · Physical consistency 60sPACE: 75.71 -> 74.94Hurt
add CGI alone to baseline for 60s generation (ablation)LongLive · Physical consistency 60sPC: 59.45 -> 62.77Improved
raise cycle loss weight from 0.1 to 0.2 (ablation)Cycle-World · VBench 5sTotal score: 84.05 -> 82.89Hurt
lower cycle loss weight from 0.1 to 0.05 (ablation)Cycle-World · Physical consistency 5sPC: 64.18 -> 56.72Hurt
spread 4 guidance steps on first timestep instead of 1 step at all 4 timesteps (ablation)CGI · Physical consistency 5sPACE: 81.91 -> 76.79Hurt
increase CGI step size from 10 to 15 (ablation)CGI · VBench 5sTotal score: 84.36 -> 83.92Mixed
use Cycle-World instead of Self Forcing for 60s generationVBench 60s · VBench 60sTotal score: 71.86 -> 82.88Improved

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