OpenCodePapers

image-generation-on-imagenet-256x256

Image Generation
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
Click a sortable column header to sort.
PaperCodeFIDInception scoreNFEModelNameReleaseDate
GenFirst: Generation Before Reconstruction for Stable End-to-End Latent Generative Modeling0.969GenFirst EiT (675M, 800 epochs)2026-08-29
SWIFT: Self-guided Weak-to-strong dIFfusion models via Target extrapolation0.99SWIFT2026-10-19
What Matters for Diffusion-Friendly Latent Manifold? Prior-Aligned Autoencoders for Latent Diffusion✓ Link1.03PAE (DINOv2)2026-05-08
HiRAE: Hierarchical Representation Autoencoding with Residual Budgets1.038HiRAE-242026-09-29
Semantics Lead the Way: Harmonizing Semantic and Texture Modeling with Asynchronous Latent Diffusion✓ Link1.04SFD (LightningDiT-XXL)2025-12-04
Unified Continuous Generative Models✓ Link1.0680SiT-XL/2 + UCGM-S (E2E-VAE + 40 sampling steps + CFG)2025-05-12
Improved Baselines with Representation Autoencoders✓ Link1.06RAEv2 (DINOv3-L, DDT-XL, K=7, REPA Guidance)2026-05-18
Hi-DiT: Hybrid Latent-Pixel Diffusion Transformer for Image Generation1.06Hi-DiT (CFG=2.4)2026-09-08
Guiding a Diffusion Transformer with the Internal Dynamics of Itself✓ Link1.07LightningDiT-XL/1 + IG2025-12-30
MixFlow Training: Alleviating Exposure Bias with Slowed Interpolation Mixture1.10MixFlow + RAE2025-12-22
Towards Scalable Pre-training of Visual Tokenizers for Generation✓ Link1.11VTP2025-12-15
An End-to-End Latent-Rollout Approach for Pushing Few-Step ImageNet-$256$ Generation to FID $1.11$ without Fréchet Losses1.11FiST (augmented)2026-09-26
Diffusion Transformers with Representation Autoencoders✓ Link1.13DiT^DH-XL (DINOv2-B)2025-10-13
RiT: Vanilla Diffusion Transformers Suffice in Representation Space✓ Link1.14RiT-XL2026-05-21
An End-to-End Latent-Rollout Approach for Pushing Few-Step ImageNet-$256$ Generation to FID $1.11$ without Fréchet Losses1.16FiST (basic)2026-09-26
CMuon: Accelerating and Stabilizing Diffusion Transformer Training via Chunked Momentum Orthogonalization1.18CMuon (DiT-XL)2026-08-03
Equivalent Flows, Unequal Learning: Clean-Latent Prediction in Transformers✓ Link1.19JLT-H/1 (L_v)2026-05-26
Direct Discriminative Optimization: Your Likelihood-Based Visual Generative Model is Secretly a GAN Discriminator✓ Link1.2150EDM2-L + DDO (SD-VAE, 25 steps, DPM-Solver-v3)2025-03-03
Unified Continuous Generative Models✓ Link1.2130UCGM-XL/2 (VA-VAE + 30 sampling steps, without guidance)2025-05-12
Unified Continuous Generative Models✓ Link1.2140UCGM-XL/2 (E2E-VAE + 40 sampling steps, without guidance)2025-05-12
Unified Continuous Generative Models✓ Link1.21100LightningDiT + UCGM-S (VA-VAE + 50 sampling steps + CFG)2025-05-12
Boosting Latent Diffusion Models via Disentangled Representation Alignment1.21Send-VAE (REPA, SiT-XL, 800 epochs)2026-01-09
MixFlow Training: Alleviating Exposure Bias with Slowed Interpolation Mixture1.22MixFlow + REPA2025-12-22
XYZFlow:Scaling Multi dimensional Shortcut Flows for Efficient Generative Modeling1.22XYZFlow-H (w/ GAN)2026-08-12
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation✓ Link1.24xAR-H2025-02-27
BitDance: Scaling Autoregressive Generative Models with Binary Tokens✓ Link1.24BitDance-H-1x2026-02-15
XYZFlow:Scaling Multi dimensional Shortcut Flows for Efficient Generative Modeling1.25XYZFlow-L (w/ GAN)2026-08-12
DDT: Decoupled Diffusion Transformer✓ Link1.26310.6DDT-XL/2(22en6de 675M + guidance interval )2025-04-08
REPA-E: Unlocking VAE for End-to-End Tuning of Latent Diffusion Transformers✓ Link1.26314.9SiT-XL/2 + REPA-E2025-04-15
SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model1.27Alitok-XL + SoftCFG2025-10-01
Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generation✓ Link1.28xAR-L2025-02-27
One Layer Is Enough: Adapting Pretrained Visual Encoders for Image Generation1.29FAE w/ Timestep Shift2025-12-08
VFM-VAE: Vision Foundation Models Can Be Good Tokenizers for Latent Diffusion Models1.31VFM-VAE2025-10-21
Condition Errors Refinement in Autoregressive Image Generation with Diffusion Loss1.31Autoregressive Condition Optimization (ACO) (943M)2026-02-02
Generative Modeling with Explicit Memory✓ Link1.32GMem (with the guidance interval)2024-12-11
FACM: Flow-Anchored Consistency Models✓ Link1.322FACM (2-step)2025-07-04
Embedding Prediction Helps Image Generation1.32NEPA-DiT-XL + REPA (SDE-250)2026-10-01
One-Step Generative Modeling via Training Dynamics Action1.33TDAction-B/22026-09-30
Diffusion Models without Classifier-free Guidance✓ Link1.34SiT-XL/2 + MG2025-02-17
Hyperspherical Latents Improve Continuous-Token Autoregressive Generation✓ Link1.34SphereAR-H2025-09-29
Revisiting Diffusion Model Predictions Through Dimensionality1.34k-Diff (LightningDiT-XL/1)2026-01-29
Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Models✓ Link1.35LightningDiT + VA-VAE (with the guidance interval)2025-01-02
AliTok: Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model✓ Link1.35318.8AliTok-XL, autoregressive, 662M2025-06-05
UDT: Reconciling U-Nets and Diffusion Transformers with Data-Adaptive Token Reduction✓ Link1.35UDT-XL/1+2026-08-02
Simpler Diffusion (SiD2): 1.5 FID on ImageNet512 with pixel-space diffusion1.38SiD22024-10-25
Rethinking Discrete Tokens: Treating Them as Conditions for Continuous Autoregressive Image Synthesis1.38DisCon-L2025-07-02
UniWeTok: An Unified Binary Tokenizer with Codebook Size $\mathit{2^{128}}$ for Unified Multimodal Large Language Model✓ Link1.38UniWeTok-H2026-02-15
Frequency-Aware Flow Matching for High-Quality Image Generation✓ Link1.38FreqFlow-H2026-04-16
SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model1.39Alitok-L + SoftCFG2025-10-01
SoftCFG: Uncertainty-guided Stable Guidance for Visual Autoregressive Model1.40Alitok-B + SoftCFG2025-10-01
U-REPA: Aligning Diffusion U-Nets to ViTs✓ Link1.41SiT↓-XL/2+U-REPA (with the guidance interval)2025-03-24
Rethinking Discrete Tokens: Treating Them as Conditions for Continuous Autoregressive Image Synthesis1.41DisCon-B2025-07-02
C^2FG: Control Classifier-Free Guidance via Score Discrepancy Analysis1.41SiT-XL/2 (REPA, Interval) + C^2FG2026-03-09
Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Think✓ Link1.42SiT-XL/2 + REPA (with the guidance interval)2024-10-09
AliTok: Towards Sequence Modeling Alignment between Tokenizer and Autoregressive Model✓ Link1.42326.6AliTok-XL, autoregressive, 318M2025-06-05
From SRA to Self-Flow: Data Augmentation or Self-Supervision?1.44SiT (XL/2) with self-representation alignment + dual-timestep scheduling + Attention Separation2026-07-02
Energy-Guided Flow Matching✓ Link1.45PixelDiT-XL + EG-FM2026-08-06
Autoregressive Visual Generation Needs a Prologue1.46Prologue L–XL2026-05-07
PixelDiT2: Representation-Grounded Pixel Diffusion Transformers✓ Link1.46PixelDiT2-H/162026-09-21
Randomized Autoregressive Visual Generation✓ Link1.48RAR-XXL, autoregressive2024-11-01
End-to-End Autoregressive Image Generation with 1D Semantic Tokenizer1.48EOSTok-H2026-05-01
Distilling Drifting Transformers with Representation Autoencoders1.48Drift-RAE (DiT DH-XL)2026-06-14
Randomized Autoregressive Visual Generation✓ Link1.50RAR-XL, autoregressive2024-11-01
SpeeDiff: Scalable Pixel-Anchored End-to-End Latent Diffusion Model1.5SpeeDiff-XL (w/ REPA++)2026-06-03
MaskBit: Embedding-free Image Generation via Bit Tokens✓ Link1.52MaskBit2024-09-24
Autoregressive Visual Generation Needs a Prologue1.52Prologue L–L2026-05-07
Generative Modeling with Explicit Memory✓ Link1.53GMem (w/o guidance)2024-12-11
REGLUE Your Latents with Global and Local Semantics for Entangled Diffusion✓ Link1.53SiT-XL/2 + REGLUE2025-12-18
Drift-AR: Single-Step Visual Autoregressive Generation via Anti-Symmetric Drifting✓ Link1.53Drift-AR (MAR-H)2026-03-30
Continuous Adversarial Flow Models1.53SiT-XL/2 + CAFM (CFG 1.3)2026-04-13
Beyond the Estimate: Wiener-Informed Decoding for Easier Training of Pixel-Space Flow Models✓ Link1.53PixelDiT-XL+WiD2026-10-26
Elucidating the design space of language models for image generation✓ Link1.54ELM2024-10-21
Hyperspherical Latents Improve Continuous-Token Autoregressive Generation✓ Link1.54SphereAR-L2025-09-29
Frequency-Aware Flow Matching for High-Quality Image Generation✓ Link1.54FreqFlow-L2026-04-16
Autoregressive Image Generation without Vector Quantization✓ Link1.55MAR-H, Diff Loss2024-06-17
Learning Representations to Guide Pixel Diffusion Transformers✓ Link1.55RG-XL/162026-10-09
PaGoDA: Progressive Growing of a One-Step Generator from a Low-Resolution Diffusion Teacher✓ Link1.56PaGoDA2024-05-23
Cross-scale Aligned Supervision for Training GANs1.56CAT-H/22026-05-26
Balancing Image Compression and Generation with Bootstrapped Tokenization1.56SelfBootTok2026-06-04
Efficient Diffusion Training via Min-SNR Weighting Strategy✓ Link1.57ViT-XL/2 with limited Interval Guidance2023-03-16
Asymmetric Flow Models✓ Link1.57AsymFlow-H/162026-05-13
Embedding Prediction Helps Image Generation1.57NEPA-DiT-XL + REPA (ODE-96)2026-10-01
MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer✓ Link1.58MDTv22023-03-25
No Other Representation Component Is Needed: Diffusion Transformers Can Provide Representation Guidance by Themselves✓ Link1.58SiT-XL + SRA2025-05-05
A Frozen Pixel-Space Diffusion Model Can Guide Itself with Its Own Samples✓ Link1.59PixelREPA-H/16 + SSG2026-07-31
Robust Latent Matters: Boosting Image Generation with Sampling Error✓ Link1.60RobustTok-L2025-03-11
PixelDiT: Pixel Diffusion Transformers for Image Generation✓ Link1.61PixelDiT-XL2025-11-25
Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models and Time-Dependent Layer Normalization✓ Link1.63DiMR-G/2R2024-06-13
PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion✓ Link1.63PixelU-H/16 (extra data)2026-06-26
Three-Body Scattering for Generative Modeling✓ Link1.63TBSM (DiT-XL/2, latent, NFE=1)2026-07-20
XYZFlow:Scaling Multi dimensional Shortcut Flows for Efficient Generative Modeling1.63XYZFlow-B (w/ GAN)2026-08-12
Persistence Forcing: Exploiting Feature Specialization in Pixel-Space Diffusion✓ Link1.63PerF-H/162026-09-28
Distribution Matching Variational AutoEncoder✓ Link1.64DMVAE (800 epochs)2025-12-08
FlowAR: Scale-wise Autoregressive Image Generation Meets Flow Matching✓ Link1.65FlowAR2024-12-19
SSG: Scaled Spatial Guidance for Multi-Scale Visual Autoregressive Generation✓ Link1.68VAR-d30 + SSG2026-02-05
UniWeTok: An Unified Binary Tokenizer with Codebook Size $\mathit{2^{128}}$ for Unified Multimodal Large Language Model✓ Link1.68UniWeTok-L2026-02-15
CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling1.70DiT-XL/2 with CADS2023-10-26
Alleviating Distortion in Image Generation via Multi-Resolution Diffusion Models and Time-Dependent Layer Normalization✓ Link1.70DiMR-XL/2R2024-06-13
Randomized Autoregressive Visual Generation✓ Link1.70RAR-L, autoregressive2024-11-01
FACM: Flow-Anchored Consistency Models✓ Link1.701FACM (1-step)2025-07-04
Dual-End Consistency Model1.7DE-CM (1 NFE)2026-02-11
FastHybrid: Accelerating Hybrid Autoregressive Image Generation with Lookahead and Guided Decoding1.7FastHybrid-H-642026-06-03
V-Co: A Closer Look at Visual Representation Alignment via Co-Denoising✓ Link1.71V-Co (H/16, DINOv2)2026-03-17
DiffiT: Diffusion Vision Transformers for Image Generation✓ Link1.73DiffiT2023-12-04
Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Prediction✓ Link1.73VAR (Visual Autoregressive)2024-04-03
End-to-End Training for Unified Tokenization and Latent Denoising✓ Link1.73UNITE-L2026-03-23
End-to-End Training for Unified Tokenization and Latent Denoising✓ Link1.75UNITE-XL2026-03-23
Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling✓ Link1.76VAR-d30 + Logit Refiner2026-09-08
Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation✓ Link1.78MAGVIT-v22023-10-09
Autoregressive Image Generation without Vector Quantization✓ Link1.78MAR-L, Diff Loss2024-06-17
MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer✓ Link1.79MDT2023-03-25
DiP: Taming Diffusion Models in Pixel Space✓ Link1.79DiP-XL/162025-11-24
WiT: Waypoint Diffusion Transformers for Alleviating Trajectory Conflict in Pixel-Space Image Generation1.79WiT-H/162026-03-16
Registers Matter for Pixel-Space Diffusion Transformers✓ Link1.8JiT-H/16 with Register Guidance + CFG2026-07-06
LDM-is-AE: Latent Diffusion Model is an Auto-Encoder for End-to-End Image Generation✓ Link1.80LDM-is-AE2026-09-29
Representation Alignment for Just Image Transformers is not Easier than You Think✓ Link1.81PixelREPA-H/16/162026-03-15
Generative Refinement Networks for Visual Synthesis✓ Link1.81GRN-G2026-04-14
HieraTok: Multi-Scale Visual Tokenizer Improves Image Reconstruction and Generation1.82HieraTok2025-09-28
Back to Basics: Let Denoising Generative Models Denoise✓ Link1.82JiT-G/162025-11-17
Improving Reconstruction of Representation Autoencoder✓ Link1.82LV-RAE (DiT DH-XL)2026-02-09
Distilling Drifting Transformers with Representation Autoencoders1.82Drift-RAE (DiT DH-L)2026-06-14
Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models✓ Link1.83Discriminator Guidance2022-11-28
Diffusion Models Need Visual Priors for Image Generation1.83DoD-XL2024-10-11
Robust Latent Matters: Boosting Image Generation with Sampling Error✓ Link1.83RobustTok-B2025-03-11
PixelGen: Pixel Diffusion Beats Latent Diffusion with Perceptual Loss✓ Link1.83PixelGen-XL/162026-02-02
MeanFlow Transformers with Representation Autoencoders✓ Link1.89MF-RAE (2 NFE) (extra data)2025-11-17
What matters for Representation Alignment: Global Information or Spatial Structure?✓ Link1.89iREPA (DINOv3-B, SiT-XL/2, 400K, with CFG)2025-12-11
OpenVision 3: A Family of Unified Visual Encoder for Both Understanding and Generation1.89OpenVision 3 (RAE)2026-01-21
REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization1.91reAR-B2025-10-06
Global Transport Couplings for Classifier-Free Guided Flows1.91SiT-XL/2 + GT2026-10-06
Hyperspherical Latents Improve Continuous-Token Autoregressive Generation✓ Link1.92SphereAR-B2025-09-29
Autoregressive Image Generation with Randomized Parallel Decoding✓ Link1.94ARPG-XXL2025-03-13
Adversarial Flow Models✓ Link1.94AF-XL/2 (4×deep, 112-layer)2025-11-27
Randomized Autoregressive Visual Generation✓ Link1.95RAR-B, autoregressive2024-11-01
Diffusion Image Generation with Explicitly Modeling of Data Manifold Geometry✓ Link1.95MIND-XL-G2026-09-08
MixTok: Adaptive Visual Tokenization with Global 1D Tokens and Sparse 2D Refinement✓ Link1.9511MixTok2026-10-09
DuoFlow: JVP-Free Finite-Difference Mean Flows for One-Step Image Generation1.96DuoFlow-XL/22026-09-08
An Image is Worth 32 Tokens for Reconstruction and Generation✓ Link1.97TiTok-S-1282024-06-11
MixFlow Training: Alleviating Exposure Bias with Slowed Interpolation Mixture1.97MixFlow + SiT-XL2025-12-22
PixelFlow: Pixel-Space Generative Models with Flow✓ Link1.98PixelFlow2025-04-10
Colored Noise Diffusion Sampling✓ Link1.98SiT-XL/2 + CNS (CFG 1.5)2026-05-28
Trajectory Forcing: Structure-First Generation with Controllable Semantic Trajectories✓ Link1.98TF-H/16 + FD-loss2026-06-21
Relay Diffusion: Unifying diffusion process across resolutions for image synthesis✓ Link1.99RDM2023-09-04
Diffusion Model is Effectively Its Own Teacher1.99PD + iSSD2025-06-11
Locality-aware Parallel Decoding for Efficient Autoregressive Image Generation✓ Link2.00LPD-XXL2025-07-02
REAR: Rethinking Visual Autoregressive Models via Generator-Tokenizer Consistency Regularization2.00reAR-S2025-10-06
ELT: Elastic Looped Transformers for Visual Generation2.0ELT-XL (7N×4L)2026-04-10
Denoising, Fast and Slow: Difficulty-Aware Adaptive Sampling for Image Generation✓ Link2.00PFT-XL/2 + REPA + Look-Ahead2026-04-21
RADAR: VQ-VAE Decoder of VAR is a Good Student for Restoring Against Degradation by Acceleration2.01VAR-d30 + RADAR2026-06-03
FasterDiT: Towards Faster Diffusion Transformers Training without Architecture Modification✓ Link2.03FasterDiT-XL/22024-10-14
Learning Stackable and Skippable LEGO Bricks for Efficient, Reconfigurable, and Variable-Resolution Diffusion Modeling✓ Link2.05338.08LEGO-XL2023-10-10
MPDiT: Multi-Patch Global-to-Local Transformer Architecture For Efficient Flow Matching and Diffusion Model✓ Link2.05MPDiT-XL2026-03-27
Vision Foundation Models as Generalist Tokenizers for Image Generation✓ Link2.05VFMTok-2B(SigLIP)2026-05-18
Rethinking Cross-Layer Information Routing in Diffusion Transformers2.05DAR dynamic c4 (ODE, CFG 1.5)2026-05-20
SiT: Exploring Flow and Diffusion-based Generative Models with Scalable Interpolant Transformers✓ Link2.06SiT-XL/2 (cfg=1.5, SDE) (extra data)2024-01-16
Next Visual Granularity Generation2.06NVG-d24 (856M+215M)2025-08-18
Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation✓ Link2.07VFMTok-3B2025-07-11
Adversarial Error Correction for Visual Autoregressive Generation✓ Link2.08AID-VAR-d242026-05-24
WiT: Waypoint Diffusion Transformers via Trajectory Conflict Navigation✓ Link2.09WiT-XL/162026-03-16
Vision Foundation Models as Generalist Tokenizers for Image Generation✓ Link2.09VFMTok-2B(DINOv2)2026-05-18
Autoregressive Image Generation with Randomized Parallel Decoding✓ Link2.1ARPG-XL2025-03-13
IAR2: Improving Autoregressive Visual Generation with Semantic-Detail Associated Token Prediction✓ Link2.10IAR2-XL2025-10-08
JSON: Jigsaw Self-play Optimization for Normalizing Flows2.1JSON2026-09-08
Training-Free Refinement of Flow Matching with Divergence-based Sampling✓ Link2.103SiT-XL + FDS2026-04-06
End-to-End Training for Unified Tokenization and Latent Denoising✓ Link2.12UNITE-B2026-03-23
Dino Forcing Flow Models: Do not denoise what you can predict2.13DF-JiT-XL/162026-10-08
SAN: Inducing Metrizability of GAN with Discriminative Normalized Linear Layer✓ Link2.14StyleSAN-XL2023-01-30
Rethinking Training Dynamics in Scale-wise Autoregressive Generation✓ Link2.14FlexVAR-d24 with SAR2025-12-06
PixNerd: Pixel Neural Field Diffusion✓ Link2.15PixNerd-XL/16 (Euler-100)2025-07-31
Autoregressive Visual Generation Needs a Prologue2.15Prologue L–B2026-05-07
Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation✓ Link2.16VFMTok-XXL(SigLIP2)2025-07-11
PixNerd: Pixel Neural Field Diffusion✓ Link2.16PixNerd-XL/16 (Adam2-50)2025-07-31
Vision Foundation Models as Generalist Tokenizers for Image Generation✓ Link2.17VFMTok-2B(SigLIP2)2026-05-18
Autoregressive Model Beats Diffusion: Llama for Scalable Image Generation✓ Link2.18LlamaGen2024-06-10
LINA: Linear Autoregressive Image Generative Models with Continuous Tokens✓ Link2.18LINA-H (cfg=2.4)2026-01-30
Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation✓ Link2.19VFMTok-XXL2025-07-11
Nested AutoRegressive Models2.22NestAR-H2025-10-27
Scalable Diffusion Models with Transformers✓ Link2.27DiT-XL/22022-12-19
PixNerd: Pixel Neural Field Diffusion✓ Link2.29PixNerd-XL/16 (Euler-50)2025-07-31
Nested AutoRegressive Models2.29NestAR-L2025-10-27
Few-Step Generation via Data-Space Iteration2.29Data-space DMD (4-NFE)2026-10-08
StyleGAN-XL: Scaling StyleGAN to Large Diverse Datasets✓ Link2.30StyleGAN-XL2022-02-01
Iwin Transformer: Hierarchical Vision Transformer using Interleaved Windows✓ Link2.30FlashDiT(win8)2025-07-24
Temporal Equilibrium MeanFlow: Bridging the Scale Gap for One-Step Generation2.3TEMF-XL+2026-06-03
Autoregressive Image Generation without Vector Quantization✓ Link2.31MAR-B, Diff Loss2024-06-17
Open-MAGVIT2: An Open-Source Project Toward Democratizing Auto-regressive Visual Generation✓ Link2.33Open-MAGVIT2-XL2024-09-06
ResTok: Learning Hierarchical Residuals in 1D Visual Tokenizers for Autoregressive Image Generation✓ Link2.34ResTok2026-01-07
Semantic One-Dimensional Tokenizer for Image Reconstruction and Generation2.34SemTok-AR-XXL2026-03-17
UniWeTok: An Unified Binary Tokenizer with Codebook Size $\mathit{2^{128}}$ for Unified Multimodal Large Language Model✓ Link2.35UniWeTok-B2026-02-15
ACDiT: Interpolating Autoregressive Conditional Modeling and Diffusion Transformer✓ Link2.37ACDiT2024-12-10
FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders✓ Link2.38FuseReg DiT-XL (p_dec=0.95, p_dit=0)2026-09-25
Improved Distributional Diffusion Models✓ Link2.38iDDM-XL/22026-09-29
Bidirectional Normalizing Flow: From Data to Noise and Back2.39BiFlow-B/22025-12-11
Vision Foundation Models as Generalist Tokenizers for Image Generation✓ Link2.41VFMTok-XL(DINOv2)2026-05-18
Autoregressive Visual Generation Needs a Prologue2.43Prologue B–XL2026-05-07
Autoregressive Image Generation with Randomized Parallel Decoding✓ Link2.44ARPG-L2025-03-13
Next Visual Granularity Generation2.44NVG-d20 (497M+125M)2025-08-18
Improved Masked Image Generation with Knowledge-Augmented Token Representations2.45AutoNAT-KA2025-11-15
An Image is Worth 32 Tokens for Reconstruction and Generation✓ Link2.48TiTok-B-642024-06-11
JLT: Clean-Latent Prediction in Latent Diffusion Transformers✓ Link2.50JLT-B/1 (guided)2026-05-26
MIMFlow: Integrating Masked Image Modeling with Normalizing Flows for End-to-End Image Generation✓ Link2.50MIMFlow-L2026-06-24
Semantic One-Dimensional Tokenizer for Image Reconstruction and Generation2.54SemTok-AR-XL2026-03-17
Adversarial Error Correction for Visual Autoregressive Generation✓ Link2.54AID-VAR-d202026-05-24
IAR2: Improving Autoregressive Visual Generation with Semantic-Detail Associated Token Prediction✓ Link2.57IAR2-L2025-10-08
Make It Efficient: Dynamic Sparse Attention for Autoregressive Image Generation2.58ADSA-3842025-06-23
GIVT: Generative Infinite-Vocabulary Transformers✓ Link2.59GIVT-Causal-L+A 2023-12-04
Hita: Holistic Tokenizer for Autoregressive Image Generation✓ Link2.59Hita-2B2025-07-03
Mean Mode Screaming: Mean--Variance Split Residuals for 1000-Layer Diffusion Transformers✓ Link2.6DiT-MV-Split (400L)2026-05-07
The Reward Was in Your Data All Along: Correcting Flow Matching with Discriminator-Guided RL2.62SiT + Discriminator-Guided RL2026-06-17
Make It Efficient: Dynamic Sparse Attention for Autoregressive Image Generation2.64ADSA-2562025-06-23
PixNerd: Pixel Neural Field Diffusion✓ Link2.64PixNerd-L/16 (Euler-50)2025-07-31
Autoregressive Visual Generation Needs a Prologue2.67Prologue B–L2026-05-07
Unlocking Any-Order Generation in Pretrained Autoregressive Image Models✓ Link2.68RAR-XL + adaptation2026-10-29
Nemotron-Labs-Diffusion-Image: Advancing Masked Discrete Diffusion for High-Resolution Image Synthesis2.69GCE (Grouped Cross-Entropy)2026-06-29
Parallel Decoding Distillation for Fast Image and Video Generation✓ Link2.69PDD (Midpoint)2026-07-28
Hita: Holistic Tokenizer for Autoregressive Image Generation✓ Link2.70Hita-XXL2025-07-03
2.74Patch Diffusion
Flow Matching for Multimodal Distributions✓ Link2.74DiTDH-XL + MM-FM (DINOv2-B, GMM Mode, AutoGuidance)2026-06-03
Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation✓ Link2.75VFMTok-L2025-07-11
An Image is Worth 32 Tokens for Reconstruction and Generation✓ Link2.77TiTok-B-322024-06-11
Semantic One-Dimensional Tokenizer for Image Reconstruction and Generation2.77SemTok-AR-L2026-03-17
Diffusion Models Need Visual Priors for Image Generation2.79DoD-B2024-10-11
Normalizing Trajectory Models2.8NTM (1.4B)2026-05-08
Efficient Training with Foresight: Multi-Token Auxiliary Supervision for Autoregressive Image Generation2.85MTAR-L2026-08-26
Polynomial Implicit Neural Representations For Large Diverse Datasets✓ Link2.86Poly-INR2023-03-20
Nested AutoRegressive Models2.86NestAR-B2025-10-27
Iwin Transformer: Hierarchical Vision Transformer using Interleaved Windows✓ Link2.90FlashDiT(win4)2025-07-24
Aligning Latent Geometry for Spherical Flow Matching in Image Generation2.91Spherical-Slerp SiT-B/2 (FLUX.2 VAE)2026-05-14
Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis✓ Link2.95RAE + Self-Flow2026-03-06
CaTok: Taming Mean Flows for One-Dimensional Causal Image Tokenization✓ Link2.95CATOK-L-1282026-03-06
Scalable GANs with Transformers2.96GAT-XL/2 + latent-space guidance2025-09-29
Hita: Holistic Tokenizer for Autoregressive Image Generation✓ Link2.98Hita-XL2025-07-03
MGVQ: Could VQ-VAE Beat VAE? A Generalizable Tokenizer with Multi-group Quantization✓ Link3.02294.1MGVQ2025-07-14
Next Visual Granularity Generation3.03NVG-d16 (255M+64M)2025-08-18
Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models✓ Link3.18ADM-G++ (FID)2022-11-28
SNCE: Geometry-Aware Supervision for Scalable Discrete Image Generation3.2IBQ-B (846M, SNCE)2026-03-16
Adversarial Error Correction for Visual Autoregressive Generation✓ Link3.24AID-VAR-d162026-05-24
Coevolving Representations in Joint Image-Feature Diffusion✓ Link3.3CoReDi-XL/2 (675M)2026-04-19
Diffusion As Self-Distillation: End-to-End Latent Diffusion In One Model3.35DSD-B2025-11-18
Stabilize the Latent Space for Image Autoregressive Modeling: A Unified Perspective✓ Link3.39205.96DiGIT-0.7B2024-10-16
Draft-and-Revise: Effective Image Generation with Contextual RQ-Transformer3.41Contextual RQ-Transformer2022-06-09
Flow Straighter and Faster: Efficient One-Step Generative Modeling via MeanFlow on Rectified Trajectories✓ Link3.41Re-Meanflow (SiT-XL, 676.7M)2025-11-28
Vision Foundation Models as Effective Visual Tokenizers for Autoregressive Image Generation✓ Link3.43VFMTok-B2025-07-11
Scaling up GANs for Text-to-Image Synthesis✓ Link3.45GigaGAN2023-03-09
BudgetPix: Compute-Adaptive Tokenization for Pixel-Space Image Diffusion✓ Link3.46BudgetPix (pMF-L/16, 60% budget)2026-10-08
Return of Unconditional Generation: A Self-supervised Representation Generation Method✓ Link3.49RCG-L (w/o guidance)2023-12-06
Speedrunning ImageNet Diffusion✓ Link3.49SR-DiT-B/1 (balanced sampling)2025-12-13
High-Resolution Image Synthesis with Latent Diffusion Models✓ Link3.60LDM-4-G2021-12-20
BIGRoC: Boosting Image Generation via a Robust Classifier✓ Link3.63BIGRoC-gt (Guided-Diffusion)2021-08-08
Language Model Beats Diffusion -- Tokenizer is Key to Visual Generation✓ Link3.65MAGVIT-v2 (w/o guidance)2023-10-09
BIGRoC: Boosting Image Generation via a Robust Classifier✓ Link3.69BIGRoC-pl (Guided-Diffusion)2021-08-08
Simple diffusion: End-to-end diffusion for high resolution images✓ Link3.71simple diffusion (U-Net)2023-01-26
Simple diffusion: End-to-end diffusion for high resolution images✓ Link3.75simple diffusion (U-ViT, L)2023-01-26
Autoregressive Image Generation using Residual Quantization✓ Link3.83RQ-Transformer2022-03-03
Diffusion Models Beat GANs on Image Synthesis✓ Link3.94ADM-G, ADM-U2021-05-11
Entropy-driven Sampling and Training Scheme for Conditional Diffusion Generation✓ Link3.96ADM-G + EDS (ED-DPM, classifier_scale=0.75)2022-06-23
MaskGIT: Masked Generative Image Transformer✓ Link4.02MaskGIT (a=0.05)2022-02-08
Image Generation with a Sphere Encoder4.02sphere-xl2026-02-16
Complexity-Balanced Diffusion Splitting✓ Link4.03SiT-XL/2 + CBS (0.4, 0.77; CFG)2026-06-04
IAR2: Improving Autoregressive Visual Generation with Semantic-Detail Associated Token Prediction✓ Link4.06IAR2-B2025-10-08
Entropy-driven Sampling and Training Scheme for Conditional Diffusion Generation✓ Link4.09ADM-G + EDS + ECT (ED-DPM, classifier_scale=1.0)2022-06-23
Autoregressive Visual Generation Needs a Prologue4.11Prologue B–B2026-05-07
Exposure Bias Can Alleviate Itself via Directional and Frequency Rectification in Flow Matching4.14DEFAR (SiT-XL/2)2026-06-26
EMAG: Self-Rectifying Diffusion Sampling with Exponential Moving Average Guidance4.16EMAG2025-12-19
Flowing Backwards: Improving Normalizing Flows via Reverse Representation Alignment✓ Link4.18Latent-TARFlow + R-REPA (Patch Size 1)2025-11-27
Learning on the Manifold: Unlocking Standard Diffusion Transformers with Representation Encoders✓ Link4.21DiT-L (459M) with RJF (DiNOv2-B)2026-02-10
4.29LDM
Adaptive Non-Uniform Timestep Sampling for Accelerating Diffusion Model Training4.33Adaptive Non-Uniform Timestep Sampling2024-11-15
Structured Residual Connectivity Matters for Diffusion Transformers4.34Fused-Mirror (SiT-XL/2 + REPA, 0.35M fine-tuning)2026-09-27
Diffusion As Self-Distillation: End-to-End Latent Diffusion In One Model4.38DSD-M2025-11-18
Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models✓ Link4.45ADM-G++ (Recall)2022-11-28
Flow Matching in Latent Space✓ Link4.46LFM2023-07-17
Residual Learning in Diffusion Models4.47Residual Learning in Diffusion Models (ADM-G)2024-06-17
LINA: Linear Autoregressive Image Generative Models with Continuous Tokens✓ Link4.49LINA-H (cfg=1.0)2026-01-30
Efficient Training with Foresight: Multi-Token Auxiliary Supervision for Autoregressive Image Generation4.50MTAR-B2026-08-26
Scalable Adaptive Computation for Iterative Generation✓ Link4.51RIN2022-12-22
Rethinking the Objectives of Vector-Quantized Tokenizers for Image Synthesis✓ Link4.55SeQ-GAN + NAR (364M)2022-12-06
Diffusion Models Beat GANs on Image Synthesis✓ Link4.59ADM-G2021-05-11
Geometry-Aware Image Flow Matching4.62SFM2026-05-24
D2PO: Optimizing Diffusion Samplers via Dynamic Preference4.7D2PO (iPNDM, 7 steps)2026-07-07
Image Generation with a Sphere Encoder4.76sphere-l2026-02-16
Scaling the Codebook Size of VQGAN to 100,000 with a Utilization Rate of 99%✓ Link4.81VQGAN-LC + LDM (32x32 tokens)2024-06-17
Cascaded Diffusion Models for High Fidelity Image Generation4.88CDM2021-05-30
COT-FM: Cluster-wise Optimal Transport Flow Matching5.11COT-FM (SiT-B/2)2026-03-11
Taming Transformers for High-Resolution Image Synthesis✓ Link5.2VQGAN+Transformer (k=600, p=1.0, a=0.05)2020-12-17
MaskGIT: Masked Generative Image Transformer✓ Link6.18MaskGIT2022-02-08
FGM-HD: Boosting Generation Diversity of Fractal Generative Models through Hausdorff Dimension Induction6.21FGM-HD2025-11-12
Taming Transformers for High-Resolution Image Synthesis✓ Link6.59VQGAN+Transformer (k=mixed, p=1.0, a=0.005)2020-12-17
ArcVQ-VAE: A Spherical Vector Quantization Framework with ArcCosine Additive Margin✓ Link6.79ArcVQ-VAE + LDM2026-05-13
Polarity Sampling: Quality and Diversity Control of Pre-Trained Generative Networks via Singular Values✓ Link6.82Polarity-BigGAN2022-03-03
Online Clustered Codebook✓ Link6.87CVQ-VAE + LDM2023-07-27
TC-Pade: Trajectory-Consistent Pade Approximation for Diffusion Acceleration6.93TC-Padé (fast)2026-03-03
MoVQ: Modulating Quantized Vectors for High-Fidelity Image Generation✓ Link7.13Mo-VQGAN-auto2022-09-19
SD-DiT: Unleashing the Power of Self-supervised Discrimination in Diffusion Transformer7.21SD-DiT-XL/22024-03-25
Empirical Variational Autoencoder✓ Link7.68EVA2026-10-05
Diffusion As Self-Distillation: End-to-End Latent Diffusion In One Model7.89DSD-S2025-11-18
Large Scale GAN Training for High Fidelity Natural Image Synthesis✓ Link8.1BigGAN-deep2018-09-28
Image Tokenizers Should Be Trained Between Images✓ Link8.48InterpTok (SiT-B/1D)2026-10-12
DiffusionBlocks: Blockwise Training for Generative Models via Score-Based Diffusion✓ Link9.0DiT-L/2 + DiffusionBlocks2025-06-17
LiFT: Loop Flow Transformers9.31LiFT XL/2 R122026-10-04
11.84ADM
Improved Denoising Diffusion Probabilistic Models✓ Link12.3Improved DDPM2021-02-18
Show the Signal, Hide the Noise: Spectral Forcing for Pixel-Space Diffusion✓ Link15.15JiT-700M/32 + Linear-SF (120 ep)2026-06-16
Trainable Log-linear Sparse Attention for Efficient Diffusion Transformers✓ Link20.41LLSA (PixelFlow)2025-12-18
ViT$^3$: Unlocking Test-Time Training in Vision✓ Link39.31DiT³-B/22025-12-01
Check out how to contribute new results. Then edit this file on GitLab or GitHub.