awesome-diffusion-categorized

Awesome Diffusion Categorized

Contents

• Visual Illusion
• Color
• Count
• Poster
• Accelerate
  • Train-Free
  • AR model
  • VAR model
• Image Restoration
  • Colorization
  • Face Restoration
  • Image Compression
  • Super Resolution
  • Personalized Restoration
• Storytelling
• Virtual Try On
• Drag Edit
• Text-Guided Editing
  • Diffusion Inversion
• Continual Learning
• Remove Concept
• In Context Learning
  • Multi-Concept
  • Decompostion
  • ID Encoder
  • General Personalization
  • AR-based
  • Video Customization

• Image Translation
• Seg & Detect & Track
• Adding Conditions
• Few-Shot
• Inpainting
• Layout

Illusion

The Art of Deception: Color Visual Illusions and Diffusion Models
[CVPR 2025] [Project] [Code]

Visual Anagrams: Generating Multi-View Optical Illusions with Diffusion Models
[CVPR 2024] [Project] [Code]

PTDiffusion: Free Lunch for Generating Optical Illusion Hidden Pictures with Phase-Transferred Diffusion Model
[CVPR 2025] [Project] [Code]

Factorized Diffusion: Perceptual Illusions by Noise Decomposition
[ECCV 2024] [Project] [Code]

Diffusion Illusions: Hiding Images in Plain Sight
[Website] [Project] [Code]

Diffusion-based Visual Anagram as Multi-task Learning
[WACV 2025] [Code]

Evaluating Model Perception of Color Illusions in Photorealistic Scenes
[Website] [Code]

Illusion3D: 3D Multiview Illusion with 2D Diffusion Priors
[Website] [Project]

Color in Generation

ColorPeel: Color Prompt Learning with Diffusion Models via Color and Shape Disentanglement
[ECCV 2024] [Project] [Code]

Free-Lunch Color-Texture Disentanglement for Stylized Image Generation
[NeurIPS 2025] [Project] [Code]

Not Every Gift Comes in Gold Paper or with a Red Ribbon: Exploring Color Perception in Text-to-Image Models
[WACV 2026] [Project] [Code]

Color Conditional Generation with Sliced Wasserstein Guidance
[NeurIPS 2025] [Code]

Leveraging Semantic Attribute Binding for Free-Lunch Color Control in Diffusion Models
[WACV 2026] [Project]

Evaluating Model Perception of Color Illusions in Photorealistic Scenes
[Website] [Code]

Exploring Palette based Color Guidance in Diffusion Models
[ACM MM 2025]

Color Me Correctly: Bridging Perceptual Color Spaces and Text Embeddings for Improved Diffusion Generation
[ACM MM 2025]

GenColorBench: A Color Evaluation Benchmark for Text-to-Image Generation Models
[Website]

Training-Free Text-Guided Color Editing with Multi-Modal Diffusion Transformer
[Website]

DiffBrush:Just Painting the Art by Your Hands
[Website]

GenColor: Generative Color-Concept Association in Visual Design
[Website]

Training-free Color-Style Disentanglement for Constrained Text-to-Image Synthesis
[Website]

Color encoding in Latent Space of Stable Diffusion Models
[Website]

Count Guidance

Make It Count: Text-to-Image Generation with an Accurate Number of Objects
[CVPR 2025] [Project] [Code]

Detection-Driven Object Count Optimization for Text-to-Image Diffusion Models
[Website] [Project] [Code]

CountCluster: Training-Free Object Quantity Guidance with Cross-Attention Map Clustering for Text-to-Image Generation
[Website] [Code]

YOLO-Count: Differentiable Object Counting for Text-to-Image Generation
[ICCV 2025]

CountDiffusion: Text-to-Image Synthesis with Training-Free Counting-Guidance Diffusion
[Website]

Poster

PosterCraft: Rethinking High-Quality Aesthetic Poster Generation in a Unified Framework
[Website] [Project] [Code]

CreatiPoster: Towards Editable and Controllable Multi-Layer Graphic Design Generation
[Website] [Code]

PosterMaker: Towards High-Quality Product Poster Generation with Accurate Text Rendering
[CVPR 2025] [Project]

POSTA: A Go-to Framework for Customized Artistic Poster Generation
[CVPR 2025] [Project]

DreamPoster: A Unified Framework for Image-Conditioned Generative Poster Design
[Website] [Project]

LogoDiffuser: Training-Free Multilingual Logo Generation and Stylization via Letter-Aware Attention Control
[Website]

Accelerate

PIXART-δ: Fast and Controllable Image Generation with Latent Consistency Models
[ICLR 2024 Spotlight] [Diffusers 1] [Diffusers 2] [Project] [Code]

SDXL-Turbo: Adversarial Diffusion Distillation
[Website] [Diffusers 1] [Diffusers 2] [Project] [Code]

Trajectory Consistency Distillation: Improved Latent Consistency Distillation by Semi-Linear Consistency Function with Trajectory Mapping
[Website] [Diffusers 1] [Diffusers 2] [Project] [Code]

LCM-LoRA: A Universal Stable-Diffusion Acceleration Module
[Website] [Diffusers] [Project] [Code]

Latent Consistency Models: Synthesizing High-Resolution Images with Few-Step Inference
[Website] [Project] [Code]

DMD2: Improved Distribution Matching Distillation for Fast Image Synthesis
[NeurIPS 2024 Oral] [Project] [Code]

DMD1: One-step Diffusion with Distribution Matching Distillation
[CVPR 2024] [Project] [Code]

Tortoise and Hare Guidance: Accelerating Diffusion Model Inference with Multirate Integration
[NeurIPS 2025] [Project] [Code]

Consistency Models
[ICML 2023] [Diffusers] [Code]

SwiftBrush: One-Step Text-to-Image Diffusion Model with Variational Score Distillation
[CVPR 2024] [Project] [Code]

SwiftBrush V2: Make Your One-Step Diffusion Model Better Than Its Teacher
[ECCV 2024] [Project] [Code]

CoDi: Conditional Diffusion Distillation for Higher-Fidelity and Faster Image Generation
[CVPR 2024] [Project] [Code]

PCM : Phased Consistency Model
[NeurIPS 2024] [Project] [Code]

Motion Consistency Model: Accelerating Video Diffusion with Disentangled Motion-Appearance Distillation
[NeurIPS 2024] [Project] [Code]

KOALA: Empirical Lessons Toward Memory-Efficient and Fast Diffusion Models for Text-to-Image Synthesis
[NeurIPS 2024] [Project] [Code]

Random Conditioning with Distillation for Data-Efficient Diffusion Model Compression
[CVPR 2025] [Project] [Code]

DIMO:Distilling Masked Diffusion Models into One-step Generator
[Website] [Project] [Code]

Flash Diffusion: Accelerating Any Conditional Diffusion Model for Few Steps Image Generation
[Website] [Project] [Code]

Timestep Embedding Tells: It's Time to Cache for Video Diffusion Model
[Website] [Project] [Code]

You Only Sample Once: Taming One-Step Text-to-Image Synthesis by Self-Cooperative Diffusion GANs
[Website] [Project] [Code]

PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Accelerator
[Website] [Project] [Code]

Scale-wise Distillation of Diffusion Models
[Website] [Project] [Code]

Simplifying, Stabilizing and Scaling Continuous-Time Consistency Models
[Website] [Project] [Code]

Adaptive Caching for Faster Video Generation with Diffusion Transformers
[Website] [Project] [Code]

FasterCache: Training-Free Video Diffusion Model Acceleration with High Quality
[Website] [Project] [Code]

Learning Few-Step Diffusion Models by Trajectory Distribution Matching
[Website] [Project] [Code]

SDXS: Real-Time One-Step Latent Diffusion Models with Image Conditions
[Website] [Project] [Code]

Reward Guided Latent Consistency Distillation
[Website] [Project] [Code]

T-Stitch: Accelerating Sampling in Pre-Trained Diffusion Models with Trajectory Stitching
[Website] [Project] [Code]

Accelerating Diffusion Sampling via Exploiting Local Transition Coherence
[Website] [Project] [Code]

AccVideo: Accelerating Video Diffusion Model with Synthetic Dataset
[Website] [Project] [Code]

One-Step Offline Distillation of Diffusion-based Models via Koopman Modeling
[Website] [Project] [Code]

Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding
[Website] [Project] [Code]

MagCache: Fast Video Generation with Magnitude-Aware Cache
[Website] [Project] [Code]

Evolutionary Caching to Accelerate Your Off-the-Shelf Diffusion Model
[Website] [Project] [Code]

Less is Enough: Training-Free Video Diffusion Acceleration via Runtime-Adaptive Caching
[Website] [Project] [Code]

Distilling Diversity and Control in Diffusion Models
[Website] [Project] [Code]

SANA-Sprint: One-Step Diffusion with Continuous-Time Consistency Distillation
[Website] [Project] [Code]

LeMiCa: Lexicographic Minimax Path Caching for Efficient Diffusion-Based Video Generation
[Website] [Project] [Code]

Glance: Accelerating Diffusion Models with 1 Sample
[Website] [Project] [Code]

No Cache Left Idle: Accelerating diffusion model via Extreme-slimming Caching
[Website] [Project] [[Code](https://github.com/THU-AccDiff/xslim/]

Relational Feature Caching for Accelerating Diffusion Transformers
[Website] [Project] [[Code](https://github.com/cvlab-yonsei]

One-Way Ticket:Time-Independent Unified Encoder for Distilling Text-to-Image Diffusion Models
[CVPR 2025]

[Code]

Relational Diffusion Distillation for Efficient Image Generation
[ACM MM 2024 (Oral)] [Code]

Autoregressive Distillation of Diffusion Transformers
[CVPR 2025 Oral] [Code]

UFOGen: You Forward Once Large Scale Text-to-Image Generation via Diffusion GANs
[CVPR 2024] [Code]

Accelerating Diffusion Transformer via Increment-Calibrated Caching with Channel-Aware Singular Value Decomposition
[CVPR 2025] [Code]

ToM: Decider-Guided Dynamic Token Merging for Accelerating Diffusion MLLMs
[AAAI 2026] [Code]

SADA: Stability-guided Adaptive Diffusion Acceleration
[ICML 2025] [Code]

SlimFlow: Training Smaller One-Step Diffusion Models with Rectified Flow
[ECCV 2024] [Code]

Accelerating Image Generation with Sub-path Linear Approximation Model
[ECCV 2024] [Code]

Diff-Instruct: A Universal Approach for Transferring Knowledge From Pre-trained Diffusion Models
[NeurIPS 2023] [Code]

Fast and Memory-Efficient Video Diffusion Using Streamlined Inference
[NeurIPS 2024] [Code]

Accelerating Diffusion via Hybrid Data-Pipeline Parallelism Based on Conditional Guidance Scheduling
[CVPR 2026] [Code]

A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models
[ICML 2024] [Code]

Score identity Distillation: Exponentially Fast Distillation of Pretrained Diffusion Models for One-Step Generation
[ICML 2024] [Code]

On the Trajectory Regularity of ODE-based Diffusion Sampling
[ICML 2024] [Code]

InstaFlow: One Step is Enough for High-Quality Diffusion-Based Text-to-Image Generation
[ICLR 2024] [Code]

Improved Training Technique for Latent Consistency Models
[ICLR 2025] [Code]

ProCache: Constraint-Aware Feature Caching with Selective Computation for Diffusion Transformer Acceleration
[AAAI 2026] [Code]

Compute Only 16 Tokens in One Timestep: Accelerating Diffusion Transformers with Cluster-Driven Feature Caching
[ACM MM 2025] [Code]

CacheQuant: Comprehensively Accelerated Diffusion Models
[CVPR 2025] [Code]

SeaCache: Spectral-Evolution-Aware Cache for Accelerating Diffusion Models
[CVPR 2026] [Code]

Accelerating Vision Diffusion Transformers with Skip Branches
[Website] [Code]

Accelerating Diffusion Transformers with Dual Feature Caching
[Website] [Code]

From Reusing to Forecasting: Accelerating Diffusion Models with TaylorSeers
[Website] [Code]

Exposure Bias Reduction for Enhancing Diffusion Transformer Feature Caching
[Website] [Code]

One Step Diffusion via Shortcut Models
[Website] [Code]

DuoDiff: Accelerating Diffusion Models with a Dual-Backbone Approach
[Website] [Code]

DraftAttention: Fast Video Diffusion via Low-Resolution Attention Guidance
[Website] [Code]

A Closer Look at Time Steps is Worthy of Triple Speed-Up for Diffusion Model Training
[Website] [Code]

Stable Consistency Tuning: Understanding and Improving Consistency Models
[Website] [Code]

SpeedUpNet: A Plug-and-Play Adapter Network for Accelerating Text-to-Image Diffusion Models
[Website] [Code]

Learning-to-Cache: Accelerating Diffusion Transformer via Layer Caching
[Website] [Code]

SDXL-Lightning: Progressive Adversarial Diffusion Distillation
[Website] [Code]

Distribution Backtracking Builds A Faster Convergence Trajectory for Diffusion Distillation
[Website] [Code]

Long and Short Guidance in Score identity Distillation for One-Step Text-to-Image Generation
[Website] [Code]

Diffusion Models Are Innate One-Step Generators
[Website] [Code]

Optimal Stepsize for Diffusion Sampling
[Website] [Code]

Model Reveals What to Cache: Profiling-Based Feature Reuse for Video Diffusion Models
[Website] [Code]

Few-Step Diffusion via Score identity Distillation
[Website] [Code]

FastCache: Fast Caching for Diffusion Transformer Through Learnable Linear Approximation
[Website] [Code]

SenseFlow: Scaling Distribution Matching for Flow-based Text-to-Image Distillation
[Website] [Code]

Sparse-vDiT: Unleashing the Power of Sparse Attention to Accelerate Video Diffusion Transformers
[Website] [Code]

Morse: Dual-Sampling for Lossless Acceleration of Diffusion Models
[Website] [Code]

SpeCa: Accelerating Diffusion Transformers with Speculative Feature Caching
[Website] [Code]

QuantSparse: Comprehensively Compressing Video Diffusion Transformer with Model Quantization and Attention Sparsification
[Website] [Code]

DC-Gen: Post-Training Diffusion Acceleration with Deeply Compressed Latent Space
[Website] [Code]

QuantSparse: Comprehensively Compressing Video Diffusion Transformer with Model Quantization and Attention Sparsification
[Website] [Code]

Towards Better & Faster Autoregressive Image Generation: From the Perspective of Entropy
[Website] [Code]

pi-Flow: Policy-Based Few-Step Generation via Imitation Distillation
[Website] [Code]

Towards One-step Causal Video Generation via Adversarial Self-Distillation
[Website] [Code]

RedVTP: Training-Free Acceleration of Diffusion Vision-Language Models Inference via Masked Token-Guided Visual Token Pruning
[Website] [Code]

Decoupled DMD: CFG Augmentation as the Spear, Distribution Matching as the Shield
[Website] [Code]

TurboDiffusion: Accelerating Video Diffusion Models by 100-200 Times
[Website] [Code]

CorGi: Contribution-Guided Block-Wise Interval Caching for Training-Free Acceleration of Diffusion Transformers
[Website] [Code]

ArcFlow: Unleashing 2-Step Text-to-Image Generation via High-Precision Non-Linear Flow Distillation
[Website] [Code]

Jano: Adaptive Diffusion Generation with Early-stage Convergence Awareness
[Website] [Code]

SODA: Sensitivity-Oriented Dynamic Acceleration for Diffusion Transformer
[Website] [Code]

TDM-R1: Reinforcing Few-Step Diffusion Models with Non-Differentiable Reward
[Website] [Code]

Distilling Diffusion Models into Conditional GANs
[ECCV 2024] [Project]

Shortcutting Pre-trained Flow Matching Diffusion Models is Almost Free Lunch
[NeurIPS 2025] [Project]

Cache Me if You Can: Accelerating Diffusion Models through Block Caching
[CVPR 2024] [Project]

Plug-and-Play Diffusion Distillation
[CVPR 2024] [Project]

SnapFusion: Text-to-Image Diffusion Model on Mobile Devices within Two Seconds
[NeurIPS 2023] [Project]

One-step Diffusion Models with f-Divergence Distribution Matching
[Website] [Project]

MagicDistillation: Weak-to-Strong Video Distillation for Large-Scale Portrait Few-Step Synthesis
[Website] [Project]

Diffusion Adversarial Post-Training for One-Step Video Generation
[Website] [Project]

SNOOPI: Supercharged One-step Diffusion Distillation with Proper Guidance
[Website] [Project]

NitroFusion: High-Fidelity Single-Step Diffusion through Dynamic Adversarial Training
[Website] [Project]

Truncated Consistency Models
[Website] [Project]

Multi-student Diffusion Distillation for Better One-step Generators
[Website] [Project]

Effortless Efficiency: Low-Cost Pruning of Diffusion Models
[Website] [Project]

SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Training
[Website] [Project]

SnapGen-V: Generating a Five-Second Video within Five Seconds on a Mobile Device
[Website] [Project]

Align Your Flow: Scaling Continuous-Time Flow Map Distillation
[Website] [Project]

Training-Free Motion Customization for Distilled Video Generators with Adaptive Test-Time Distillation
[Website] [Project]

Forecasting When to Forecast: Accelerating Diffusion Models with Confidence-Gated Taylor
[Website] [Project]

POSE: Phased One-Step Adversarial Equilibrium for Video Diffusion Models
[Website] [Project]

Hyper-Bagel: A Unified Acceleration Framework for Multimodal Understanding and Generation
[Website] [Project]

Large Scale Diffusion Distillation via Score-Regularized Continuous-Time Consistency
[Website] [Project]

Self-Evaluation Unlocks Any-Step Text-to-Image Generation
[Website] [Project]

Transition Matching Distillation for Fast Video Generation
[Website] [Project]

Fast Autoregressive Video Diffusion and World Models with Temporal Cache Compression and Sparse Attention
[Website] [Project]

FlashBlock: Attention Caching for Efficient Long-Context Block Diffusion
[Website] [Project]

Adversarial Distribution Matching for Diffusion Distillation Towards Efficient Image and Video Synthesis
[ICCV 2025 (Highlight)]

OmniCache: A Trajectory-Oriented Global Perspective on Training-Free Cache Reuse for Diffusion Transformer Models
[ICCV 2025]

FasterDiT: Towards Faster Diffusion Transformers Training without Architecture Modification
[NeurIPS 2024]

One-Step Diffusion Distillation through Score Implicit Matching
[NeurIPS 2024]

Self-Corrected Flow Distillation for Consistent One-Step and Few-Step Text-to-Image Generation
[AAAI 2025]

TAP: A Token-Adaptive Predictor Framework for Training-Free Diffusion Acceleration
[CVPR 2026]

BlockDance: Reuse Structurally Similar Spatio-Temporal Features to Accelerate Diffusion Transformers
[CVPR 2025]

PCM : Picard Consistency Model for Fast Parallel Sampling of Diffusion Models
[CVPR 2025]

MVPortrait: Text-Guided Motion and Emotion Control for Multi-view Vivid Portrait Animation
[CVPR 2025]

Revisiting Diffusion Models: From Generative Pre-training to One-Step Generation
[ICML 2025]

Accelerate High-Quality Diffusion Models with Inner Loop Feedback
[Website]

Accelerating Diffusion Transformer via Error-Optimized Cache
[Website]

DICE: Distilling Classifier-Free Guidance into Text Embeddings
[Website]

ProReflow: Progressive Reflow with Decomposed Velocity
[Website]

Inference-Time Diffusion Model Distillation
[Website]

Taming Consistency Distillation for Accelerated Human Image Animation
[Website]

Token Pruning for Caching Better: 9 Times Acceleration on Stable Diffusion for Free
[Website]

HarmoniCa: Harmonizing Training and Inference for Better Feature Cache in Diffusion Transformer Acceleration
[Website]

Diff-Instruct*: Towards Human-Preferred One-step Text-to-image Generative Models
[Website]

MLCM: Multistep Consistency Distillation of Latent Diffusion Model
[Website]

EM Distillation for One-step Diffusion Models
[Website]

AsymRnR: Video Diffusion Transformers Acceleration with Asymmetric Reduction and Restoration
[Website]

Score-of-Mixture Training: Training One-Step Generative Models Made Simple
[Website]

Partially Conditioned Patch Parallelism for Accelerated Diffusion Model Inference
[Website]

Importance-based Token Merging for Diffusion Models
[Website]

Imagine Flash: Accelerating Emu Diffusion Models with Backward Distillation
[Website]

Accelerating Diffusion Models with One-to-Many Knowledge Distillation
[Website]

Accelerating Video Diffusion Models via Distribution Matching
[Website]

TDDSR: Single-Step Diffusion with Two Discriminators for Super Resolution
[Website]

DDIL: Improved Diffusion Distillation With Imitation Learning
[Website]

OSV: One Step is Enough for High-Quality Image to Video Generation
[Website]

Target-Driven Distillation: Consistency Distillation with Target Timestep Selection and Decoupled Guidance
[Website]

Token Caching for Diffusion Transformer Acceleration
[Website]

DiP-GO: A Diffusion Pruner via Few-step Gradient Optimization
[Website]

LazyDiT: Lazy Learning for the Acceleration of Diffusion Transformers
[Website]

Flow Generator Matching
[Website]

Multistep Distillation of Diffusion Models via Moment Matching
[Website]

SFDDM: Single-fold Distillation for Diffusion models
[Website]

LAPTOP-Diff: Layer Pruning and Normalized Distillation for Compressing Diffusion Models
[Website]

CogView3: Finer and Faster Text-to-Image Generation via Relay Diffusion
[Website]

SCott: Accelerating Diffusion Models with Stochastic Consistency Distillation
[Website]

Ditto: Accelerating Diffusion Model via Temporal Value Similarity
[Website]

Adaptive Non-Uniform Timestep Sampling for Diffusion Model Training
[Website]

TSD-SR: One-Step Diffusion with Target Score Distillation for Real-World Image Super-Resolution
[Website]

Sparse VideoGen: Accelerating Video Diffusion Transformers with Spatial-Temporal Sparsity
[Website]

Efficient Distillation of Classifier-Free Guidance using Adapters
[Website]

Denoising Score Distillation: From Noisy Diffusion Pretraining to One-Step High-Quality Generation
[Website]

Inductive Moment Matching
[Website]

High Quality Diffusion Distillation on a Single GPU with Relative and Absolute Position Matching
[Website]

DiTFastAttnV2: Head-wise Attention Compression for Multi-Modality Diffusion Transformers
[Website]

Mean Flows for One-step Generative Modeling
[Website]

Faster Video Diffusion with Trainable Sparse Attention
[Website]

Accelerating Diffusion-based Super-Resolution with Dynamic Time-Spatial Sampling
[Website]

SRDiffusion: Accelerate Video Diffusion Inference via Sketching-Rendering Cooperation
[Website]

Sparse VideoGen2: Accelerate Video Generation with Sparse Attention via Semantic-Aware Permutation
[Website]

RainFusion: Adaptive Video Generation Acceleration via Multi-Dimensional Visual Redundancy
[Website]

Foresight: Adaptive Layer Reuse for Accelerated and High-Quality Text-to-Video Generation
[Website]

Accelerating Diffusion Large Language Models with SlowFast: The Three Golden Principles
[Website]

Diffusion Transformer-to-Mamba Distillation for High-Resolution Image Generation
[Website]

Upsample What Matters: Region-Adaptive Latent Sampling for Accelerated Diffusion Transformers
[Website]

Accelerating Parallel Diffusion Model Serving with Residual Compression
[Website]

SwiftVideo: A Unified Framework for Few-Step Video Generation through Trajectory-Distribution Alignment
[Website]

MixCache: Mixture-of-Cache for Video Diffusion Transformer Acceleration
[Website]

Forecast then Calibrate: Feature Caching as ODE for Efficient Diffusion Transformers
[Website]

DiCache: Let Diffusion Model Determine Its Own Cache
[Website]

HiCache: Training-free Acceleration of Diffusion Models via Hermite Polynomial-based Feature Caching
[Website]

SpecDiff: Accelerating Diffusion Model Inference with Self-Speculation
[Website]

BWCache: Accelerating Video Diffusion Transformers through Block-Wise Caching
[Website]

RAPID^3: Tri-Level Reinforced Acceleration Policies for Diffusion Transformer
[Website]

SLA: Beyond Sparsity in Diffusion Transformers via Fine-Tunable Sparse-Linear Attention
[Website]

CLQ: Cross-Layer Guided Orthogonal-based Quantization for Diffusion Transformers
[Website]

Score Distillation of Flow Matching Models
[Website]

Let Features Decide Their Own Solvers: Hybrid Feature Caching for Diffusion Transformers
[Website]

LinVideo: A Post-Training Framework towards O(n) Attention in Efficient Video Generation
[Website]

FreqCa: Accelerating Diffusion Models via Frequency-Aware Caching
[Website]

Hierarchical Koopman Diffusion: Fast Generation with Interpretable Diffusion Trajectory
[Website]

Test-Time Iterative Error Correction for Efficient Diffusion Models
[Website]

From Structure to Detail: Hierarchical Distillation for Efficient Diffusion Model
[Website]

PipeDiT: Accelerating Diffusion Transformers in Video Generation with Task Pipelining and Model Decoupling
[Website]

Flash-DMD: Towards High-Fidelity Few-Step Image Generation with Efficient Distillation and Joint Reinforcement Learning
[Website]

GalaxyDiT: Efficient Video Generation with Guidance Alignment and Adaptive Proxy in Diffusion Transformers
[Website]

InvarDiff: Cross-Scale Invariance Caching for Accelerated Diffusion Models
[Website]

USV: Unified Sparsification for Accelerating Video Diffusion Models
[Website]

TwinFlow: Realizing One-step Generation on Large Models with Self-adversarial Flows
[Website]

Few-Step Distillation for Text-to-Image Generation: A Practical Guide
[Website]

On the Design of One-step Diffusion via Shortcutting Flow Paths
[Website]

OUSAC: Optimized Guidance Scheduling with Adaptive Caching for DiT Acceleration
[Website]

Plug-and-Play Fidelity Optimization for Diffusion Transformer Acceleration via Cumulative Error Minimization
[Website]

Forecast the Principal, Stabilize the Residual: Subspace-Aware Feature Caching for Efficient Diffusion Transformers
[Website]

DisCa: Accelerating Video Diffusion Transformers with Distillation-Compatible Learnable Feature Caching
[Website]

NanoFLUX: Distillation-Driven Compression of Large Text-to-Image Generation Models for Mobile Devices
[Website]

DDiT: Dynamic Patch Scheduling for Efficient Diffusion Transformers
[Website]

LESA: Learnable Stage-Aware Predictors for Diffusion Model Acceleration
[Website]

Analyzing and Improving Fast Sampling of Text-to-Image Diffusion Models
[Website]

Adaptive Spectral Feature Forecasting for Diffusion Sampling Acceleration
[Website]

TC-Padé: Trajectory-Consistent Padé Approximation for Diffusion Acceleration
[Website]

Train-Free

AsyncDiff: Parallelizing Diffusion Models by Asynchronous Denoising
[NeurIPS 2024] [Project] [Code]

Training-Free Adaptive Diffusion with Bounded Difference Approximation Strategy
[NeurIPS 2024] [Project] [Code]

DeepCache: Accelerating Diffusion Models for Free
[CVPR 2024] [Project] [Code]

Grouping First, Attending Smartly: Training-Free Acceleration for Diffusion Transformers
[Website] [Project] [Code]

Faster Diffusion: Rethinking the Role of the Encoder for Diffusion Model Inference
[NeurIPS 2024] [Code]

DiTFastAttn: Attention Compression for Diffusion Transformer Models
[NeurIPS 2024] [Code]

Structural Pruning for Diffusion Models
[NeurIPS 2023] [Code]

AutoDiffusion: Training-Free Optimization of Time Steps and Architectures for Automated Diffusion Model Acceleration
[ICCV 2023] [Code]

Agent Attention: On the Integration of Softmax and Linear Attention
[ECCV 2024] [Code]

Attend to Not Attended: Structure-then-Detail Token Merging for Post-training DiT Acceleration
[CVPR 2025] [Code]

Token Merging for Fast Stable Diffusion
[CVPRW 2024] [Code]

LightCache: Memory-Efficient, Training-Free Acceleration for Video Generation
[Website] [Code]

FORA: Fast-Forward Caching in Diffusion Transformer Acceleration
[Website] [Code]

Real-Time Video Generation with Pyramid Attention Broadcast
[Website] [Code]

Accelerating Diffusion Transformers with Token-wise Feature Caching
[Website] [Code]

TGATE-V1: Cross-Attention Makes Inference Cumbersome in Text-to-Image Diffusion Models
[Website] [Code]

TGATE-V2: Faster Diffusion via Temporal Attention Decomposition
[Website] [Code]

SmoothCache: A Universal Inference Acceleration Technique for Diffusion Transformers
[Website] [Code]

Attention-Driven Training-Free Efficiency Enhancement of Diffusion Models
[CVPR 2024] [Project]

Training-free Diffusion Acceleration with Bottleneck Sampling
[Website] [Project]

Cache Me if You Can: Accelerating Diffusion Models through Block Caching
[Website] [Project]

Fewer Denoising Steps or Cheaper Per-Step Inference: Towards Compute-Optimal Diffusion Model Deployment
[ICCV 2025]

Token Fusion: Bridging the Gap between Token Pruning and Token Merging
[WACV 2024]

Flexiffusion: Training-Free Segment-Wise Neural Architecture Search for Efficient Diffusion Models
[Website]

PFDiff: Training-free Acceleration of Diffusion Models through the Gradient Guidance of Past and Future
[Website]

Δ-DiT: A Training-Free Acceleration Method Tailored for Diffusion Transformers
[Website]

Adversarial Score identity Distillation: Rapidly Surpassing the Teacher in One Step
[Website]

Diff-Instruct++: Training One-step Text-to-image Generator Model to Align with Human Preferences
[Website]

Fast constrained sampling in pre-trained diffusion models
[Website]

Chipmunk: Training-Free Acceleration of Diffusion Transformers with Dynamic Column-Sparse Deltas
[Website]

ETC: training-free diffusion models acceleration with Error-aware Trend Consistency
[Website]

AR model

Distilled Decoding 1: One-step Sampling of Image Auto-regressive Models with Flow Matching
[ICLR 2025] [Project] [Code]

Accelerating Auto-regressive Text-to-Image Generation with Training-free Speculative Jacobi Decoding
[ICLR 2025] [Code]

LANTERN: Accelerating Visual Autoregressive Models with Relaxed Speculative Decoding
[ICLR 2025] [Code]

Show-o Turbo: Towards Accelerated Unified Multimodal Understanding and Generation
[Website] [Code]

SimpleAR: Pushing the Frontier of Autoregressive Visual Generation through Pretraining, SFT, and RL
[Website] [Code]

Speculative Jacobi-Denoising Decoding for Accelerating Autoregressive Text-to-image Generation
[Website]

SJD++: Improved Speculative Jacobi Decoding for Training-free Acceleration of Discrete Auto-regressive Text-to-Image Generation
[Website]

Hawk: Leveraging Spatial Context for Faster Autoregressive Text-to-Image Generation
[Website]

Fast-ARDiff: An Entropy-informed Acceleration Framework for Continuous Space Autoregressive Generation
[Website]

VAR model

Collaborative Decoding Makes Visual Auto-Regressive Modeling Efficient
[CVPR 2025] [Project] [Code]

FastVAR: Linear Visual Autoregressive Modeling via Cached Token Pruning
[ICCV 2025] [Project] [Code]

Memory-Efficient Visual Autoregressive Modeling with Scale-Aware KV Cache Compression
[Website] [Code]

SkipVAR: Accelerating Visual Autoregressive Modeling via Adaptive Frequency-Aware Skipping
[Website] [Code]

Frequency-Aware Autoregressive Modeling for Efficient High-Resolution Image Synthesis
[Website] [Code]

LiteVAR: Compressing Visual Autoregressive Modelling with Efficient Attention and Quantization
[Website]

Image Restoration

Diffusion Prior-Based Amortized Variational Inference for Noisy Inverse Problems
[ECCV 2024 Oral] [Project] [Code]

Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model
[ICLR 2023 oral] [Project] [Code]

Scaling Up to Excellence: Practicing Model Scaling for Photo-Realistic Image Restoration In the Wild
[CVPR 2024] [Project] [Code]

Selective Hourglass Mapping for Universal Image Restoration Based on Diffusion Model
[CVPR 2024] [Project] [Code]

Zero-Reference Low-Light Enhancement via Physical Quadruple Priors
[CVPR 2024] [Project] [Code]

From Posterior Sampling to Meaningful Diversity in Image Restoration
[ICLR 2024] [Project] [Code]

Generative Diffusion Prior for Unified Image Restoration and Enhancement
[CVPR 2023] [Project] [Code]

MoE-DiffIR: Task-customized Diffusion Priors for Universal Compressed Image Restoration
[ECCV 2024] [Project] [Code]

Image Restoration with Mean-Reverting Stochastic Differential Equations
[ICML 2023] [Project] [Code]

PhoCoLens: Photorealistic and Consistent Reconstruction in Lensless Imaging
[NeurIPS 2024 Spotlight] [Project] [Code]

Denoising Diffusion Models for Plug-and-Play Image Restoration
[CVPR 2023 Workshop NTIRE] [Project] [Code]

FoundIR: Unleashing Million-scale Training Data to Advance Foundation Models for Image Restoration
[Website] [Project] [Code]

Improving Diffusion Inverse Problem Solving with Decoupled Noise Annealing
[Website] [Project] [Code]

SVFR: A Unified Framework for Generalized Video Face Restoration
[Website] [Project] [Code]

DiffIR2VR-Zero: Zero-Shot Video Restoration with Diffusion-based Image Restoration Models
[Website] [Project] [Code]

Solving Video Inverse Problems Using Image Diffusion Models
[Website] [Project] [Code]

RestoreVAR: Visual Autoregressive Generation for All-in-One Image Restoration
[Website] [Project] [Code]

Learning Efficient and Effective Trajectories for Differential Equation-based Image Restoration
[Website] [Project] [Code]

GenDR: Lightning Generative Detail Restorator
[Website] [Project] [Code]

AutoDIR: Automatic All-in-One Image Restoration with Latent Diffusion
[Website] [Project] [Code]

SeedVR2: One-Step Video Restoration via Diffusion Adversarial Post-Training
[Website] [Project] [Code]

Text-Aware Image Restoration with Diffusion Models
[Website] [[Proj