Autoregressive video synthesis offers a promising pathway for infinite-horizon generation but is fundamentally hindered by three intertwined challenges: semantic forgetting from context limitations, visual drift due to positional extrapolation, and controllability loss during interactive instruction switching. Current methods often tackle these issues in isolation, limiting long-term coherence. We introduce Grounded Forcing, a novel framework that bridges time-independent semantics and proximal dynamics through three interlocking mechanisms. First, to address semantic forgetting, we propose a Dual Memory KV Cache that decouples local temporal dynamics from global semantic anchors, ensuring long-term semantic coherence and identity stability. Second, to suppress visual drift, we design Dual-Reference RoPE Injection, which confines positional embeddings within the training manifold while rendering global semantics time-invariant. Third, to resolve controllability issues, we develop Asymmetric Proximity Recache, which facilitates smooth semantic inheritance during prompt transitions via proximity-weighted cache updates. These components operate synergistically to tether the generative process to stable semantic cores while accommodating flexible local dynamics. Extensive experiments demonstrate that Grounded Forcing significantly enhances long-range consistency and visual stability, establishing a robust foundation for interactive long-form video synthesis.
Figures & tables
Figure 1: Grounded Forcing for Long-Horizon Interactive Video Generation. Our method generates a coherent 1-minute video with multiple characters (Grandma, cat, and dog) across diverse scenes, maintaining long-term identity consistency while supporting dynamic prompt switching and multi-shot transitions.
Figure 2: Limitations of Existing Methods. (a) Infinity-RoPE in Multi-Shot Generation: The model remains fixated on the initial frame (man, red box) and fails to preserve newly introduced entities (Hulk in yellow, Silver Wyvern in purple), demonstrating semantic forgetting . (b) LongLive in Continuous Generation: Rigid anchoring to initial semantics causes visual inconsistencies during semantic transitions, hindering semantic evolution and controllability.
Figure 3: Dual Memory Mechanism. (a) Conventional KV Cache with a single sink frame, which suffers from semantic dilution over time. (b) Our Dual Memory KV Cache separates Local Temporal Memory (sliding window) from Global Consistency Memory (persistent anchors). (c) Case of Global Memory Update: Upon prompt switching, the global memory is selectively updated based on latent diversity. New frames with high semantic novelty (low similarity to GCM) replace redundant anchors (high similarity to remaining GCM), ensuring the memory spans the evolving semantic space (e.g., from Hulk to Hulk and Silver Wyvern).
Figure 4: Dual-Reference RoPE Injection. Visualization of the positional encoding management over generation steps. Orange blocks represent Global Consistency Memory (GCM) keys, always injected with a fixed RoPE index of 0 to ensure position-agnostic semantic anchoring. Blue blocks denote Local Temporal Memory (LTM) keys, assigned relative indices within the model’s perceptual range (e.g., [0,21] ). Gray blocks indicate evicted frames, labeled with their absolute temporal positions (e.g., 2,…,i−1 ) to illustrate the continuous timeline extending beyond the fixed cache window. This strategy ensures indices remain within the training distribution while preserving long-term semantics.
Figure 5: Asymmetric Proximity Recache (APR). Comparison of KV Cache strategies during prompt switching. (a-b) Standard caching fails to adapt. (c) Uniform KV ReCache refreshes all frames equally, causing semantic discontinuity. (d) Our APR applies a proximity-dependent recache scale (arrow), refreshing recent frames more aggressively to follow the new prompt while retaining distant frames to preserve semantic inheritance.
Model
Results on 240s ↑
Aesthetic Quality
Background Consistency
Dynamic Degree
Imaging Quality
Motion Smoothness
Subject Consistency
Temporal Flickering
LongLive
0.6032
0.9208
0.60
0.7245
0.9892
0.8973
0.9770
Rolling Forcing
0.6237
0.9144
0.53
0.7317
0.9878
0.9034
0.9845
Infinity Rope
0.6048
0.9195
0.58
0.7154
0.9873
0.9066
0.9808
Ours
0.6174
0.9265
0.60
0.7204
0.9878
0.9163
0.9812
Table 1: Performance comparisons on 240s videos. We evaluate seven metrics covering aesthetic quality, temporal consistency, and motion dynamics. Best and second-best results are bolded and underlined , respectively.
Figure 6: Qualitative Comparison with Baselines. (a) Multi-Shot Generation: Our method maintains consistent identity through multiple semantic transitions (man → Hulk → Silver Wyvern), while Infinity-RoPE suffers from identity drift. (b) Single-Shot Generation: Our approach preserves Iron Man’s identity and action coherence (laser → explosion), whereas LongLive and Infinity-RoPE exhibit visual degradation.
Model
Ablation Results on 60s ↑
Aesthetic Quality
Background Consistency
Subject Consistency
Temporal Flickering
w/o Dual Mem
0.6473
0.8467
0.7322
0.9734
w/o DR-RoPE
0.6278
0.8729
0.7606
0.9716
w/o APR
0.6386
0.8603
0.7719
0.9636
Ours (full)
0.6440
0.8795
0.7770
0.9723
Table 2: Ablation study on component effectiveness. We evaluate the impact of removing key modules (Dual Mem, DR-RoPE, APR) on 60s multi-prompt generation. Best and second-best results are bolded and underlined , respectively.
Metric
LongLive
Infinity-Rope
Ours
Subject Consistency
3.02
2.98
3.66
Background Consistency
3.04
3.08
3.78
Aesthetic Quality
3.45
3.34
3.42
Text Adherence
3.12
3.15
3.72
Table 3: User study results on 5-point Likert scale. We compare our method with LongLive and Infinity-Rope across four perceptual metrics. Higher scores indicate better performance. Best results are in bold .
Appendix figures & tables6 assets
Supplementary material from the paper’s appendix.
Appendix
Model
Results on 5s ↑
Aesthetic Quality
Background Consistency
Dynamic Degree
Imaging Quality
Motion Smoothness
Subject Consistency
Temporal Flickering
LongLive
0.6512
0.9534
0.62
0.7823
0.9934
0.9412
0.9923
Rolling Forcing
0.6689
0.9487
0.58
0.7891
0.9928
0.9387
0.9931
Infinity Rope
0.6445
0.9501
0.60
0.7756
0.9921
0.9445
0.9918
Ours
0.6623
0.9612
0.60
0.7801
0.9924
0.9421
0.9931
Appendix
Table 4: Performance comparisons on 5s videos. All methods perform well at short horizons; our approach maintains advantages in semantic consistency metrics. Best and second-best results are bolded and underlined , respectively.
Model
Results on 60s ↑
Aesthetic Quality
Background Consistency
Dynamic Degree
Imaging Quality
Motion Smoothness
Subject Consistency
Temporal Flickering
LongLive
0.6145
0.9289
0.61
0.7534
0.9912
0.9123
0.9834
Rolling Forcing
0.6378
0.9267
0.56
0.7612
0.9901
0.9256
0.9867
Infinity Rope
0.6189
0.9313
0.59
0.7445
0.9895
0.9234
0.9845
Ours
0.6298
0.9356
0.61
0.7489
0.9900
0.9317
0.9841
Appendix
Table 5: Performance comparisons on 60s videos. Our method shows growing advantages in consistency metrics as generation horizon extends. Best and second-best results are bolded and underlined , respectively.
Figure 7: Multi-shot generation with narrative continuity (60s). Character identity (highlighted in red boxes) remains consistent across diverse scenes and camera angles. The GCM successfully anchors core semantic features while allowing local dynamics to adapt to new prompts via APR and LTM Reset.
Figure 8: Single-prompt generation (240s). Our method demonstrates stronger subject-matter and background consistency in long-range generation results compared to other methods.
Figure 9: Interactive Prompt Switching (60s). Our method demonstrates stronger consistency of main content and smoother video transitions in long-range interactive prompt generation results.
Figure 10: Interactive Prompt Switching (60s). Our method demonstrates stronger consistency of main content and smoother video transitions in long-range interactive prompt generation results.