mlsanityData to 7 Oct 2026

Infinity-RoPE: Action-Controllable Infinite Video Generation Emerges From Autoregressive Self-Rollout

arXiv 2511.20649 ↗First posted 25 Nov 2025cs.CVcited by 78

On the map: Diffusion models → Video generation

Abstract (arXiv)

Current autoregressive video diffusion models are constrained by three core bottlenecks: (i) the finite temporal horizon imposed by the base model's 3D Rotary Positional Embedding (3D-RoPE), (ii) slow prompt responsiveness in maintaining fine-grained action control during long-form rollouts, and (iii) the inability to realize discontinuous cinematic transitions within a single generation stream. We introduce ∞-RoPE, a unified inference-time framework that addresses all three limitations through three interconnected components: Block-Relativistic RoPE, KV Flush, and RoPE Cut. Block-Relativistic RoPE reformulates temporal encoding as a moving local reference frame, where each newly generated latent block is rotated relative to the base model's maximum frame horizon while earlier blocks are rotated backward to preserve relative temporal geometry. This relativistic formulation eliminates fixed temporal positions, enabling continuous video generation far beyond the base positional limits. To obtain fine-grained action control without re-encoding, KV Flush renews the KV cache by retaining only two latent frames, the global sink and the last generated latent frame, thereby ensuring immediate prompt responsiveness. Finally, RoPE Cut introduces controlled discontinuities in temporal RoPE coordinates, enabling multi-cut scene transitions within a single continuous rollout. Together, these components establish ∞-RoPE as a training-free foundation for infinite-horizon, controllable, and cinematic video diffusion. Comprehensive experiments show that ∞-RoPE consistently surpasses previous autoregressive models in overall VBench scores.

Key facts. Cited by 78 papers in the map (data through 7 Oct 2026). Placed in Video generation, one of 14 directions in Diffusion models. Builds on 34 of its 41 references that are in the map. First posted on arXiv 25 Nov 2025.

Builds on

34 of 41 references.

Closest ancestors · 16

Same direction, newest first

  1. 2025Stable Video Infinity: Infinite-Length Video Generation with Error Recycling10 Oct
  2. 2025Self-Forcing++: Towards Minute-Scale High-Quality Video Generation2 Oct
  3. 2025Rolling Forcing: Autoregressive Long Video Diffusion in Real Time29 Sep
  4. 2025StreamDiT: Real-Time Streaming Text-to-Video Generation4 Jul
  5. 2025Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion9 Jun
  6. 2025SkyReels-V2: Infinite-length Film Generative Model17 Apr
  7. 2024LTX-Video: Realtime Video Latent Diffusion30 Dec
  8. 2024Autoregressive Video Generation without Vector Quantization18 Dec
  9. 2024From Slow Bidirectional to Fast Autoregressive Video Diffusion Models10 Dec
  10. 2024Pyramidal Flow Matching for Efficient Video Generative Modeling8 Oct
  11. 2024CogVideoX: Text-to-Video Diffusion Models with An Expert Transformer12 Aug
  12. 2023Photorealistic Video Generation with Diffusion Models11 Dec
4 more
  1. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets25 Nov
  2. 2023Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models18 Apr
  3. 2022Imagen Video: High Definition Video Generation with Diffusion Models5 Oct
  4. 2022Video Diffusion Models7 Apr

Other references · 18

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2020Denoising Diffusion Implicit Modelscited by 9,655
  3. 2022Scalable Diffusion Models with Transformerscited by 5,931
  4. 2022Flow Matching for Generative Modelingcited by 5,426
  5. 2021RoFormer: Enhanced Transformer with Rotary Position Embeddingcited by 4,581
  6. 2024Scaling Rectified Flow Transformers for High-Resolution Image Synthesiscited by 4,094
  7. 2025Wan: Open and Advanced Large-Scale Video Generative Modelscited by 2,536
  8. 2023VBench: Comprehensive Benchmark Suite for Video Generative Modelscited by 1,755
  9. 2022Make-A-Video: Text-to-Video Generation without Text-Video Datacited by 1,725
  10. 2024HunyuanVideo: A Systematic Framework For Large Video Generative Modelscited by 1,597
  11. 2023One-step Diffusion with Distribution Matching Distillationcited by 978
  12. 2022CogVideo: Large-scale Pretraining for Text-to-Video Generation via Transformerscited by 973
6 more
  1. 2024Improved Distribution Matching Distillation for Fast Image Synthesiscited by 648
  2. 2024Diffusion Forcing: Next-token Prediction Meets Full-Sequence Diffusioncited by 634
  3. 2024Movie Gen: A Cast of Media Foundation Modelscited by 614
  4. 2025MAGI-1: Autoregressive Video Generation at Scalecited by 262
  5. 2025LongLive: Real-time Interactive Long Video Generationcited by 237
  6. 2019Scaling Autoregressive Video Modelscited by 189

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 5; 2026: 73.

78 papers.

Sort:

By area:

  1. 2026Causal Forcing: Autoregressive Diffusion Distillation Done Right for High-Quality Real-Time Interactive Video Generationcited by 159
  2. 2025WorldPlay: Towards Long-Term Geometric Consistency for Real-Time Interactive World Modelingcited by 152
  3. 2025Reward Forcing: Efficient Streaming Video Generation with Rewarded Distribution Matching Distillationcited by 75
  4. 2026Context Forcing: Consistent Autoregressive Video Generation with Long Contextcited by 47
  5. 2026Infinite Worlds with Versatile Interactionscited by 37
  6. 2026DreamX-World 1.0: A General-Purpose Interactive World Modelcited by 33
  7. 2026LoL: Longer than Longer, Scaling Video Generation to Hourcited by 29
  8. 2026Rolling Sink: Bridging Limited-Horizon Training and Open-Ended Testing in Autoregressive Video Diffusioncited by 28
  9. 2026Causal-rCM: A Unified Teacher-Forcing and Self-Forcing Open Recipe for Autoregressive Diffusion Distillation in Streaming Video Generation and Interactive World Modelscited by 23
  10. 2026ShotStream: Streaming Multi-Shot Video Generation for Interactive Storytellingcited by 22
  11. 2026MemRoPE: Training-Free Infinite Video Generation via Evolving Memory Tokenscited by 20
  12. 2026Relax Forcing: Relaxed KV-Memory for Consistent Long Video Generationcited by 20
38 more
  1. 2026Self Gradient Forcing: Native Long Video Extrapolationcited by 16
  2. 2026MBench: A Comprehensive Benchmark on Memory Capability for Video World Modelscited by 15
  3. 2026Anchor Forcing: Anchor Memory and Tri-Region RoPE for Interactive Streaming Video Diffusioncited by 14
  4. 2026One-Forcing: Towards Stable One-Step Autoregressive Video Generationcited by 13
  5. 2026Pathwise Test-Time Correction for Autoregressive Long Video Generationcited by 12
  6. 2026Mode Seeking meets Mean Seeking for Fast Long Video Generationcited by 12
  7. 2026LongLive-RAG: A General Retrieval-Augmented Framework for Long Video Generationcited by 12
  8. 2026OPSD-V: On-Policy Self-Distillation for Post-Training Few-Step Autoregressive Video Generatorscited by 12
  9. 2026Salt: Self-Consistent Distribution Matching with Cache-Aware Training for Fast Video Generationcited by 11
  10. 2026Addressable Memory for Video World Modelscited by 11
  11. 2026Forcing-KV: Hybrid KV Cache Compression for Efficient Autoregressive Video Diffusion Modelscited by 10
  12. 2025BIFE: Better Interaction, Fewer Errors for Minute-Long Video Generationcited by 9
  13. 2025LongVie 2: Multimodal Controllable Ultra-Long Video World Modelcited by 9
  14. 2026Train Short, Inference Long: Training-free Horizon Extension for Autoregressive Video Generationcited by 9
  15. 2026CausalCine: Real-Time Autoregressive Generation for Multi-Shot Video Narrativescited by 8
  16. 2026Long-Horizon Streaming Video Generation via Hybrid Attention with Decoupled Distillationcited by 7
  17. 2026TetherCache: Stabilizing Autoregressive Long-Form Video Generation with Gated Recall and Trusted Alignmentcited by 7
  18. 2026Efficient Video Diffusion Models: Advancements and Challengescited by 6
  19. 2026RAVEN: Real-time Autoregressive Video Extrapolation with Consistency-model GRPOcited by 6
  20. 2026EntityBench: Towards Entity-Consistent Long-Range Multi-Shot Video Generationcited by 6
  21. 2026Echo-Infinity: Learning Evolving Memory for Real-Time Infinite Video Generationcited by 6
  22. 2026Video Generation Models as World Models: Efficient Paradigms, Architectures and Algorithmscited by 5
  23. 2026Grounded Forcing: Bridging Time-Independent Semantics and Proximal Dynamics in Autoregressive Video Synthesiscited by 5
  24. 2026Head Forcing: Long Autoregressive Video Generation via Head Heterogeneitycited by 5
  25. 2026FadeMem: Distance-Aware Memory Consolidation for Autoregressive Video Diffusioncited by 5
  26. 2026Context-Matched Distillation: Teacher Causality for Autoregressive Video Distillationcited by 5
  27. 2026Ring Forcing: Towards Precise Long-Term Memory for Autoregressive Video Diffusioncited by 5
  28. 2026Pyramid Forcing: Head-Aware Pyramid KV Cache Policy for High-Quality Long Video Generationcited by 4
  29. 2026UniTemp: Unlocking Video Generation in Any Temporal Order via Bidirectional Distillationcited by 4
  30. 2026Vidu S1: A Real-Time Interactive Video Generation Modelcited by 4
  31. 2026Flex-Forcing: Towards a Unified Autoregressive and Bidirectional Video Diffusion Modelcited by 4
  32. 2026StreamAV-Bench: A Comprehensive Benchmark for Streaming Audio-Video Generationcited by 4
  33. 2025JoyStreamer-Flash: Real-time and Infinite Audio-Driven Avatar Generation with Autoregressive Diffusioncited by 3
  34. 2026FashionChameleon: Towards Real-Time and Interactive Human-Garment Video Customizationcited by 3
  35. 2026VideoMLA: Low-Rank Latent KV Cache for Minute-Scale Autoregressive Video Diffusioncited by 3
  36. 2026FreqForcing: Autoregressive Long Video Generation via Spectral Self-Anchoringcited by 3
  37. 2026DCARL: A Divide-and-Conquer Framework for Autoregressive Long-Trajectory Video Generationcited by 2
  38. 2026World-Ego Modeling for Long-Horizon Evolution in Hybrid Embodied Taskscited by 2

50 most cited of 78 All 78 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025VideoSSM: Autoregressive Long Video Generation with Hybrid State-Space Memory
  2. 2025Playing with Transformer at 30+ FPS via Next-Frame Diffusion
  3. 2026Directing the World: Fast Autoregressive Video Generation with Compositional Human-Camera Control
  4. 2026SPIRAL: Self-Evolving Action-Conditioned Video Generation via Reflective Planning Agents
  5. 2026Accelerating Video Generation Inference with Sequential-Parallel 3D Positional Encoding Using a Global Time Index
  6. 2026TempAct: Advancing Temporal Plausibility in Autoregressive Video Generation via Planner-Executor RL
  7. 2025AR-Diffusion: Asynchronous Video Generation with Auto-Regressive Diffusion
  8. 2026SoulX-LiveAct: Towards Hour-Scale Real-Time Human Animation with Neighbor Forcing and ConvKV Memory