j1s4nn/jev-adaptive-granularity-figures

JevGAG Adaptive Temporal Granularity Figures

Publication figures and frame grids for adaptive temporal granularity

JevGAG Adaptive Temporal Granularity Figures

Result figures, qualitative frame grids, and contact sheets for the study "Jev-Routed Temporal Granularity with Recovery and Hysteresis for Causal Video Generation".

The study audits three preconditions for adaptive temporal granularity in causal video generation on Causal-rCM (Wan2.1-T2V-1.3B): (P1) a mode that is genuinely cheaper on the deployment hardware, (P2) a signal available before generation that predicts hard segments, and (P3) an endpoint independent of the modes being compared. Frame-wise (c1-1), chunk-3 (c3-3), and a near-zero-cost hold mode are routed by a two-threshold controller with hysteresis and forced recovery, driven by the Jev System One oracle. Audits run on a Tesla T4 and an A100-SXM4-80GB.

Contents

• fig_*.png / fig_*.pdf: manuscript figures (cost-quality Pareto on T4 and A100, hardware scaling and break-even, signal AUROC, Jev routing traces, controller ablation, monotonicity, confirmatory bootstrap, A100/T4 online policies, hold-mode study).
• architecture.png/.pdf: system diagram of pre-generation routing and causal rollout.
• qualitative_grid.png/.pdf: matched decoded frames across policies.
• qual_*.png: individual qualitative start/end frames for fixed fine and fixed coarse rollouts.
• contact_*.jpg: contact sheets of decoded frames across prompts.

Companion datasets

• Result videos: https://huggingface.co/datasets/j1s4nn/jev-adaptive-granularity-videos
• Audited results (E1-E9): https://huggingface.co/datasets/j1s4nn/jevgag-results

Related resources

• Code (JevGAG programme): https://github.com/j1s4nn/JevGAG
• Code (Jev-Routed Temporal Granularity): https://github.com/j1s4nn/jev-adaptive-granularity-WM
• Project page (JevGAG): https://j1s4nn.github.io/JevGAG/
• Project page (Paper 1): https://j1s4nn.github.io/jev-adaptive-granularity-WM/
• Portfolio: https://j1s4nn.github.io
• Source (Kaggle): https://www.kaggle.com/datasets/ajjisan/jev-adaptive-granularity-figures
• Preprint (JevGAG): https://doi.org/10.5281/zenodo.23693517
• Preprint (Jev-Routed Temporal Granularity): https://doi.org/10.5281/zenodo.23395414

Citation

@misc{hossen_jevgag_ddag,
  author       = {Hossen, MD Jisan},
  title        = {Regime-Aware Temporal Granularity for Causal Video World Models: When Does Adaptivity Pay?},
  year         = {2026},
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.23693517},
  url          = {https://doi.org/10.5281/zenodo.23693517}
}

License

CC-BY-4.0. Maintained by Vision Intelligence (MD Jisan Hossen (NUIST)).