Research map / Video / Video generation
Deep Forcing: Training-Free Long Video Generation with Deep Sink and Participative Compression
Jung Yi, Wooseok Jang, Paul Hyunbin Cho, Jisu Nam, Heeji Yoon, Seungryong Kim
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Deep Sink to Self Forcing (ablation) | Self Forcing · VBench-Long 30s | Dynamic Degree: 36.62 -> 48.58 | Improved |
| add Participative Compression on top of Deep Sink (ablation) | SF + Deep Sink · VBench-Long 30s | Dynamic Degree: 48.58 -> 57.56 | Improved |
| add Participative Compression on top of Deep Sink (ablation) | SF + Deep Sink · VBench-Long 30s | Image Quality: 68.54 -> 69.31 | Improved |
| add Deep Sink to Self Forcing (ablation) | Self Forcing · VBench-Long 30s | Image Quality: 68.58 -> 68.54 | No clear effect |
| add Deep Sink and Participative Compression to Self Forcing (ablation) | Self Forcing · VBench-Long 30s | Overall Consistency: 20.50 -> 20.54 | Improved |
| select Top-C with both past and denoising queries instead of only denoising (ablation) | Participative Compression · VBench-Long 30s | Image Quality: 68.24 -> 69.31 | Improved |
| select Top-C with both past and denoising queries instead of only past (ablation) | Participative Compression · VBench-Long 30s | Image Quality: 68.54 -> 69.31 | Improved |
| select Top-C with both queries instead of only denoising (ablation) | Participative Compression · VBench-Long 30s | Motion Smoothness: 97.86 -> 98.27 | Improved |
| apply Deep Forcing training-free to Self Forcing | Self Forcing · VBench-Long 60s | Dynamic Degree: 31.98 -> 57.19 | Improved |
| apply Deep Forcing training-free to Self Forcing | Self Forcing · VBench-Long 60s | Imaging Quality: 66.33 -> 69.27 | Improved |
| apply Deep Forcing to Self Forcing | Self Forcing · VLM visual stability | visual stability: 43.94 -> 75.44 | Improved |
| add Deep Forcing compression to Self Forcing | Self Forcing · 60s generation, H100 | throughput FPS: 15.78 -> 15.75 | No clear effect |
Later papers that compared with this method
7 later accepted papers report Deep Forcing as a baseline in their results tables; 7 of them report a better number on most of the results they share with it.
- PackForcing: Short Video Training Suffices for Long Video Sampling and Long Context Inference (ECCV 2026 · better on 5 of 6 shared results)
- Learning-to-Memorize: Dynamic Context Management for Long-Horizon Autoregressive Video Generation (NeurIPS 2026 · better on 5 of 5 shared results)
- What Remains in Sight? Autoregressive Video Decoding as Representation-Guided Context Rewriting (NeurIPS 2026 · better on 3 of 3 shared results)
- Disen-Forcing: Disentangling Semantic Anchoring from Motion for Autoregressive Video Diffusion (NeurIPS 2026 · better on 3 of 3 shared results)
- Advancing Narrative Long Video Generation via Training-Free Identity-Aware Memory (NeurIPS 2026 · better on 2 of 2 shared results)
- VideoMLA: Low-Rank Latent KV Cache for Minute-Scale Autoregressive Video Diffusion (NeurIPS 2026 · better on 2 of 2 shared results)
- Anchor Forcing: Anchor Memory and Tri-Region RoPE for Interactive Streaming Video Diffusion (ECCV 2026 · better on 2 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 2 on most shared results.
- Rolling Forcing — Rolling Forcing: Autoregressive Long Video Diffusion in Real Time (ICLR 2026) · better on 3 of 6 shared results
- LongLive — LongLive: Real-time Interactive Long Video Generation (ICLR 2026) · better on 5 of 6 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
- Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion (NeurIPS 2025 · similarity 0.91)
- Disen-Forcing: Disentangling Semantic Anchoring from Motion for Autoregressive Video Diffusion (NeurIPS 2026 · similarity 0.91)
- Flex-Forcing: Towards a Unified Autoregressive and Bidirectional Video Diffusion Model (ICML 2026 · similarity 0.91)
- Rolling Forcing: Autoregressive Long Video Diffusion in Real Time (ICLR 2026 · similarity 0.91)
- Self-Forcing++: Towards Minute-Scale High-Quality Video Generation (ICLR 2026 · similarity 0.91)
- From Slow Bidirectional to Fast Autoregressive Video Diffusion Models (CVPR 2025 · similarity 0.90)
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.
