Awesome Efficient Video Generation

A curated list of recent efficient video generation methods.

🗂️ Table of Contents

• Long Context
• Auto-Regressive
• Attention
• Caching
• VAE
• Hierarchical
• Pruning
• Architecture
• Distillation
• Quantization
• Parallelism
• Scheduler

📄 Papers

Long Context

• Memory Forcing: Spatio-Temporal Memory for Consistent Scene Generation on Minecraft
  2025.10 | Paper

• VideoNSA: Native Sparse Attention Scales Video Understanding
  2025.10 | Paper | Code |

• Pack and Force Your Memory: Long-form and Consistent Video Generation
  2025.10 | Paper

• EvoWorld: Evolving Panoramic World Generation with Explicit 3D Memory
  2025.10 | Paper | Code |

• LongLive: Real-time Interactive Long Video Generation
  2025.09 | Paper | Code |

• Longscape: Advancing Long-horizon Embodied World Models with Context-aware MoE
  2025.09 | Paper | Code |

• KeyWorld: Key Frame Reasoning Enables Effective and Efficient World Models
  2025.09 | paper | code

• Kling-Avatar: Grounding Multimodal Instructions for Cascaded Long-Duration Avatar Animation Synthesis
  2025.09 | Paper

• HunyuanWorld-Voyager: Technical Report
  2025.08 | Paper | Code |

• Mixture of Contexts for Long Video Generation
  2025.08 | Paper

• Macro-from-Micro Planning for High-Quality and Parallelized Autoregressive Long Video Generation
  2025.08 | Paper | Code |

• LoViC: Efficient Long Video Generation with Context Compression
  2025.07 | Paper

• Video World Models with Long-term Spatial Memory
  2025.06 | Paper

• Context as Memory: Scene-Consistent Interactive Long Video Generation with Memory Retrieval
  2025.06 | Paper

• Memory-Efficient Visual Autoregressive Modeling with Scale-Aware KV Cache Compression
  2025.05 | Paper | Code |

• Packing Input Frame Context in Next-Frame Prediction Models for Video Generation
  2025.04 | Paper | Code |

• Long-Context Autoregressive Video Modeling with Next-Frame Prediction
  2025.03 | Paper | Code |

Auto-Regressive

• Longscape: Advancing Long-horizon Embodied World Models with Context-aware MoE
  2025.09 | Paper | Code |

• Macro-from-Micro Planning for High-Quality and Parallelized Autoregressive Long Video Generation
  2025.08 | Paper | Code |

• Autoregressive Adversarial Post-Training for Real-Time Interactive Video Generation
  2025.06 | Paper

• Self Forcing: Bridging the Train-Test Gap in Autoregressive Video Diffusion
  2025.06 | Paper | Code |

• Memory-Efficient Visual Autoregressive Modeling with Scale-Aware KV Cache Compression
  2025.05 | Paper | Code |

• Packing Input Frame Context in Next-Frame Prediction Models for Video Generation
  2025.04 | Paper | Code |

• From Slow Bidirectional to Fast Autoregressive Video Diffusion Models
  2024.12 | Paper | Code |

Attention

• Input-Aware Sparse Attention for Real-Time Co-Speech Video Generation
  2025.10 | Paper

• SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse–Linear Attention
  2025.09 | Paper | Code |

• LongLive: Real-time Interactive Long Video Generation
  2025.09 | Paper | Code |

• RAPID^3: Tri-Level Reinforced Acceleration Policies for Diffusion Transformer
  2025.09 | Paper

• Towards Interpretable and Efficient Attention: Compressing All by Contracting a Few
  2025.09 | Paper

• FG-Attn: Leveraging Fine-Grained Sparsity In Diffusion Transformers
  2025.09 | Paper

• Bidirectional Sparse Attention for Faster Video Diffusion Training
  2025.09 | Paper

• Compact Attention: Exploiting Structured Spatio-Temporal Sparsity for Fast Video Generation
  2025.08 | Paper

• Video-BLADE: Block-Sparse Attention Meets Step Distillation for Efficient Video Generation
  2025.08 | Paper | Code |

• VMoBA: Mixture-of-Block Attention for Video Diffusion Models
  2025.06 | Paper | Code |

• Radial Attention: O(nlogn) Sparse Attention with Energy Decay for Long Video Generation
  2025.06 | Paper | Code |

• Astraea: A GPU-Oriented Token-wise Acceleration Framework for Video Diffusion Transformers
  2025.06 | Paper

• FPSAttention: Training-Aware FP8 and Sparsity Co-Design for Fast Video Diffusion
  2025.06 | Paper

• Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
  2025.06 | Paper

• Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation
  2025.05 | Paper

• VORTA: Efficient Video Diffusion via Routing Sparse Attention
  2025.05 | Paper | Code |

• Faster Video Diffusion with Trainable Sparse Attention
  2025.05 | Paper | Code |

• Training-free and Adaptive Sparse Attention for Efficient Long Video Generation
  2025.02 | Paper

• MoBA: Mixture of Block Attention for Long-Context LLMs
  2025.02 | Paper | Code |

• Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention
  2025.02 | Paper

• Fast Video Generation with Sliding Tile Attention
  2025.02 | Paper | Code |

• Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity
  2025.02 | Paper | Code |

• AsymRnR: Video Diffusion Transformers Acceleration with Asymmetric Reduction and Restoration
  2024.12 | Paper | Code |

Caching

• RAPID^3: Tri-Level Reinforced Acceleration Policies for Diffusion Transformer
  2025.09 | Paper

• SpecDiff: Accelerating Diffusion Model Inference with Self-Speculation
  2025.09 | Paper

• BWCache: Accelerating Video Diffusion Transformers through Block-Wise Caching
  2025.09 | Paper

• SpeCa: Accelerating Diffusion Transformers with Speculative Feature Caching
  2025.09 | Paper | Code |

• Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching
  2025.09 | Paper | Code |

• ERTACache: Error Rectification and Timesteps Adjustment for Efficient Diffusion
  2025.08 | Paper

• Reusing Computation in Text-to-Image Diffusion for Efficient Generation of Image Sets
  2025.08 | Paper

• DiCache: Let Diffusion Model Determine Its Own Cache
  2025.08 | Paper | Code |

• HiCache: Training-free Acceleration of Diffusion Models via Hermite Polynomial-based Feature Caching
  2025.08 | Paper

• OmniCache: A Trajectory-Oriented Global Perspective on Training-Free Cache Reuse for Diffusion Transformer Models
  2025.08 | Paper

• Forecast then Calibrate: Feature Caching as ODE for Efficient Diffusion Transformers
  2025.08 | Paper

• MixCache: Mixture-of-Cache for Video Diffusion Transformer Acceleration
  2025.08 | Paper

• TaoCache: Structure-Maintained Video Generation Acceleration
  2025.08 | Paper

• Sortblock: Similarity-Aware Feature Reuse for Diffusion Model
  2025.08 | Paper | Code |

• PromptTea: Let Prompts Tell TeaCache the Optimal Threshold
  2025.07 | Paper | Code |

• Less is Enough: Training-Free Video Diffusion Acceleration via Runtime-Adaptive Caching
  2024.07 | Paper | Code |

• MagCache: Fast Video Generation with Magnitude-Aware Cache
  2025.06 | Paper | Code |

• FastCache: Fast Caching for Diffusion Transformer Through Learnable Linear Approximation
  2025.05 | Paper | Code |

• AB-Cache: Training-Free Acceleration of Diffusion Models via Adams-Bashforth Cached Feature Reuse
  2025.04 | Paper

• Model Reveals What to Cache: Profiling-Based Feature Reuse for Video Diffusion Models
  2025.04 | Paper | Code |

• FEB-Cache: Frequency-Guided Exposure Bias Reduction for Enhancing Diffusion Transformer Caching
  2025.03 | Paper

• From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeers
  2025.03 | Paper | Code |

• QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation
  2025.03 | Paper | Code |

• UniCP: A Unified Caching and Pruning Framework for Efficient Video Generation
  2025.02 | Paper

• Accelerating Diffusion Transformer via Error-Optimized Cache
  2025.01 | Paper

• Accelerating Diffusion Transformers with Dual Feature Caching
  2024.12 | Paper | Code |

• Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model
  2024.11 | Paper | Code |

• SmoothCache: A Universal Inference Acceleration Technique for Diffusion Transformers
  2024.11 | Paper

• Adaptive Caching for Faster Video Generation with Diffusion Transformers
  2024.11 | Paper | Code |

• FasterCache: Training-Free Video Diffusion Model Acceleration with High Quality
  2024.10 | Paper | Code |

• Accelerating Diffusion Transformers with Token-wise Feature Caching
  2024.10 | Paper | Code |

• Real-Time Video Generation with Pyramid Attention Broadcast
  2024.08 | Paper | Code |

• ∆-DiT: A Training-Free Acceleration Method Tailored for Diffusion Transformers
  2024.06 | Paper

• DiTFastAttn: Attention Compression for Diffusion Transformer Models
  2024.06 | Paper | Code |

• Faster Diffusion via Temporal Attention Decomposition
  2024.04 | Paper | Code |

VAE

• DC-VideoGen: Efficient Video Generation with Deep Compression Video Autoencoder
  2025.09 | Paper | Code |

• DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space
  2025.09 | Paper | Code |

• REGEN: Learning Compact Video Embedding with (Re-)Generative Decoder
  2025.03 | Paper

Hierarchical

• Waver: Wave Your Way to Lifelike Video Generation
  2025.08 | Paper | Code |

• SRDiffusion: Accelerate Video Diffusion Inference via Sketching-Rendering Cooperation
  2025.05 | Paper

• VGDFR: Diffusion-based Video Generation with Dynamic Latent Frame Rate
  2025.04 | Paper | Code |

• Training-free Diffusion Acceleration with Bottleneck Sampling
  2025.03 | Paper

• TPDiff: Temporal Pyramid Video Diffusion Model
  2025.03 | Paper | Code |

• DLFR-VAE: Dynamic Latent Frame Rate VAE for Video Generation
  2025.02 | Paper

• Collaborative Decoding Makes Visual Auto-Regressive Modeling Efficient
  2024.11 | Paper | Code |

• Pyramidal Flow Matching for Efficient Video Generative Modeling
  2024.10 | Paper | Code |

Pruning

• SlimDiff: Training-Free, Activation-Guided Hands-free Slimming of Diffusion Models
  2025.09 | Paper

• Do Sparse Subnetworks Exhibit Cognitively Aligned Attention? Effects of Pruning on Saliency Map Fidelity, Sparsity, and Concept Coherence
  2025.09 | Paper

• HERO: Hierarchical Extrapolation and Refresh for Efficient World Models
  2025.08 | Paper

• TokensGen: Harnessing Condensed Tokens for Long Video Generation
  2025.07 | Paper | Code |

• Astraea: A GPU-Oriented Token-wise Acceleration Framework for Video Diffusion Transformers
  2025.06 | Paper

• D2iT: Dynamic Diffusion Transformer for Accurate Image Generation
  2025.04 | Paper

• FlexiDiT: Your Diffusion Transformer Can Easily Generate High-Quality Samples with Less Compute
  2025.02 | Paper

• UniCP: A Unified Caching and Pruning Framework for Efficient Video Generation
  2025.02 | Paper

• AsymRnR: Video Diffusion Transformers Acceleration with Asymmetric Reduction and Restoration
  2024.12 | Paper | Code |

• TinyFusion: Diffusion Transformers Learned Shallow
  2024.12 | Paper | Code |

• Importance-Based Token Merging for Efficient Image and Video Generation
  2024.11 | Paper

• Token Merging for Fast Stable Diffusion
  2023.03 | Paper | Code |

Architecture

• PatchVSR: Breaking Video Diffusion Resolution Limits with Patch-wise Video Super-Resolution
  2025.09 | Paper

• HunyuanImage 3.0 Technical Report
  2025.09 | Paper | Code |

• Aligning Visual Foundation Encoders to Tokenizers for Diffusion Models
  2025.09 | Paper

• Improving Video Diffusion Transformer Training by Multi-Feature Fusion and Alignment from Self-Supervised Vision Encoders
  2025.09 | Paper

• Test-Time Training Done Right
  2025.05 | Paper | Code |

• One-Minute Video Generation with Test-Time Training
  2025.04 | Paper | Code |

Distillation

• Post-Training Quantization via Residual Truncation and Zero Suppression for Diffusion Models
  2025.09 | Paper

• Score Distillation of Flow Matching Models
  2025.09 | Paper

• DistillKac: Few-Step Image Generation via Damped Wave Equations
  2025.09 | Paper

• SD3.5-Flash: Distribution-Guided Distillation of Generative Flows
  2025.09 | Paper

• Q-Sched: Pushing the Boundaries of Few-Step Diffusion Models with Quantization-Aware Scheduling
  2025.09 | Paper

• POSE: Phased One-Step Adversarial Equilibrium for Video Diffusion Models
  2025.08 | Paper

• Ouroboros: Single-step Diffusion Models for Cycle-consistent Forward and Inverse Rendering
  2025.08 | Paper

• Dual-Expert Consistency Model for Efficient and High-Quality Video Generation
  2025.06 | Paper | Code |

• QuantCache: Adaptive Importance-Guided Quantization with Hierarchical Latent and Layer Caching for Video Generation
  2025.03 | Paper | Code |

• Distilling Diffusion Models into Conditional GANs
  2024.05 | Paper

• Improved Distribution Matching Distillation for Fast Image Synthesis
  2024.05 | Paper | Code |

• Fast High-Resolution Image Synthesis with Latent Adversarial Diffusion Distillation
  2024.03 | Paper

• LCM-LoRA: A Universal Stable-Diffusion Acceleration Module
  2023.11 | Paper | Code |

• One-step Diffusion with Distribution Matching Distillation
  2023.11 | Paper

• Adversarial Diffusion Distillation
  2023.11 | Paper

Quantization

• Q-Sched: Pushing the Boundaries of Few-Step Diffusion Models with Quantization-Aware Scheduling
  2025.09 | Paper

• FPSAttention: Training-Aware FP8 and Sparsity Co-Design for Fast Video Diffusion
  2025.06 | Paper

• SVDQuant: Absorbing Outliers by Low-Rank Components for 4-Bit Diffusion Models
  2024.11 | Paper | Code |

• ViDiT-Q: Efficient and Accurate Quantization of Diffusion Transformers for Image and Video Generation
  2024.06 | Paper | Code |

Parallelism

• STADI: Fine-Grained Step-Patch Diffusion Parallelism for Heterogeneous GPUs
  2025.09 | Paper

• KnapFormer: An Online Load Balancer for Efficient Diffusion Transformers Training
  2025.08 | Paper | Code |

• DDiT: Dynamic Resource Allocation for Diffusion Transformer Model Serving
  2025.06 | Paper

• xDiT: an Inference Engine for Diffusion Transformers (DiTs) with Massive Parallelism
  2024.05 | Paper | Code |

• PipeFusion: Patch-level Pipeline Parallelism for Diffusion Transformers Inference
  2024.05 | Paper | Code |

• DSP: Dynamic Sequence Parallelism for Multi-Dimensional Transformers
  2024.03 | Paper | Code |

• DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models
  2024.02 | Paper | Code |

Scheduler

• EVODiff: Entropy-aware Variance Optimized Diffusion Inference
  2024.05 | Paper | Code |

• ContextFlow: Training-Free Video Object Editing via Adaptive Context Enrichment
  2025.09 | Paper

• A-FloPS: Accelerating Diffusion Sampling with Adaptive Flow Path Sampler
  2025.09 | Paper

• Pretrained Diffusion Models Are Inherently Skipped-Step Samplers
  2025.08 | Paper

• Disentanglement in T-space for Faster and Distributed Training of Diffusion Models with Fewer Latent-states
  2025.08 | Paper

• Pyramidal Flow Matching for Efficient Video Generative Modeling
  2024.10 | Paper | Code |

• AsyncDiff: Parallelizing Diffusion Models by Asynchronous Denoising
  2024.06 | Paper | Code |

• AdaDiff: Adaptive Step Selection for Fast Diffusion Models
  2023.11 | Paper

• Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference
  2023.10 | Paper | Code |

• DPM-Solver++: Fast Solver for Guided Sampling of Diffusion Probabilistic Models
  2022.11 | Paper | Code |

• Flow Matching for Generative Modeling
  2022.10 | Paper

• Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow
  2022.09 | Paper | Code |

• DPM-Solver: A Fast ODE Solver for Diffusion Probabilistic Model Sampling in Around 10 Steps
  2022.06 | Paper | Code |

• Denoising Diffusion Implicit Models
  2020.10 | Paper | Code |