ICLR 2026 accepted papers on Image generation
476 of the 5,358 papers accepted at ICLR 2026 (8.9%) are on Image generation, compared with 9.3% at ICLR 2025. They fall into 7 topics.
See ICLR 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Fast diffusion sampling · 201 papers
- Enhancing Diffusion-Based Sampling with Molecular Collective Variables (1 citation)
- SPREAD: Sampling-based Pareto front Refinement via Efficient Adaptive Diffusion (1 citation)
- A Noise is Worth Diffusion Guidance
- A Unification of Discrete, Gaussian, and Simplicial Diffusion
- Adaptive Moments are Surprisingly Effective for Plug-and-Play Diffusion Sampling
- Alternating Diffusion for Proximal Sampling with Zeroth Order Queries
- Antithetic Noise in Diffusion Models
- Any-Order Flexible Length Masked Diffusion
- Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Models
- Beyond Masks: Efficient, Flexible Diffusion Language Models via Deletion-Insertion Processes
- Beyond Scattered Acceptance: Fast and Coherent Inference for DLMs via Longest Stable Prefixes
- Block-wise Adaptive Caching for Accelerating Diffusion Policy
and 189 more in the interactive map.
Autoregressive and latent generation · 107 papers
- OXtal: An All-Atom Diffusion Model for Organic Crystal Structure Prediction (2 citations)
- MVAR: Visual Autoregressive Modeling with Scale and Spatial Markovian Conditioning (1 citation)
- Steering Autoregressive Music Generation with Recursive Feature Machines (1 citation)
- YuE: Scaling Open Foundation Models for Long-Form Music Generation (1 citation)
- A Hidden Semantic Bottleneck in Conditional Embeddings of Diffusion Transformers
- Adapting Self-Supervised Representations as a Latent Space for Efficient Generation
- Align Your Structures: Generating Trajectories with Structure Pretraining for Molecular Dynamics
- Attack-Resistant Watermarking for AIGC Image Forensics via Diffusion-based Semantic Deflection
- Autoregressive Image Generation with Randomized Parallel Decoding
- Autoregressive Visual Decoding from EEG Signals
- Charts Are Not Images: On the Challenges of Scientific Chart Editing
- CIAR: Interval-based Collaborative Decoding for Image Generation Acceleration
and 95 more in the interactive map.
Text-to-image diffusion · 97 papers
- Story-Iter: A Training-free Iterative Paradigm for Long Story Visualization (3 citations)
- Implicit Inversion turns CLIP into a Decoder (1 citation)
- $\alpha$-DPO: Robust Preference Alignment for Diffusion Models via $\alpha$ Divergence
- Asynchronous Denoising Diffusion Models for Aligning Text-to-Image Generation
- AutoFigure: Generating and Refining Publication-Ready Scientific Illustrations
- BideDPO: Conditional Image Generation with Simultaneous Text and Condition Alignment
- BranchGRPO: Stable and Efficient GRPO with Structured Branching in Diffusion Models
- CASteer: Cross-Attention Steering for Controllable Concept Erasure
- CoDi: Subject-Consistent and Pose-Diverse Text-to-Image Generation
- Continual Unlearning for Text-to-Image Diffusion Models: A Regularization Perspective
- Culture in Action: Evaluating Text-to-Image Models through Social Activities
- DeLeaker: Dynamic Inference-Time Reweighting For Semantic Leakage Mitigation in Text-to-Image Models
and 85 more in the interactive map.
Editing and style transfer · 27 papers
- Adaptive Domain Shift in Diffusion Models for Cross-Modality Image Translation
- ArtUV: Artist-style UV Unwrapping
- Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?
- CoCoDiff: Correspondence-Consistent Diffusion Model for Fine-grained Style Transfer
- Constantly Improving Image Models Need Constantly Improving Benchmarks
- Detective SAM: Adaptive AI-Image Forgery Localization
- DiffVax: Optimization-Free Image Immunization Against Diffusion-Based Editing
- Does FLUX Already Know How to Perform Physically Plausible Image Composition?
- DragFlow: Unleashing DiT Priors with Region-Based Supervision for Drag Editing
- Geometric Image Editing via Effects-Sensitive In-Context Inpainting with Diffusion Transformers
- Inverse Virtual Try-On: Generating Multi-Category Product-Style Images from Clothed Individuals
- Optimizing ID Consistency in Multimodal Large Models: Facial Restoration via Alignment, Entanglement, and Disentanglement
and 15 more in the interactive map.
Super-resolution · 21 papers
- Breaking Scale Anchoring: Frequency Representation Learning for Accurate High-Resolution Inference from Low-Resolution Training
- Bridging Degradation Discrimination and Generation for Universal Image Restoration
- Eliminating VAE for Fast and High-Resolution Generative Detail Restoration
- Improved Adversarial Diffusion Compression for Real-World Video Super-Resolution
- LinearSR: Unlocking Linear Attention for Stable and Efficient Image Super-Resolution
- SRT: Super-Resolution for Time Series via Disentangled Rectified Flow
- Step-Aware Residual-Guided Diffusion for EEG Spatial Super-Resolution
- The False Promise of Zero-Shot Super-Resolution in Machine-Learned Operators
- Trust but Verify: Adaptive Conditioning for Reference-Based Diffusion Super-Resolution via Implicit Reference Correlation Modeling
- VARestorer: One-Step VAR Distillation for Real-World Image Super-Resolution
- Bridging the Distribution Gap to Harness Pretrained Diffusion Priors for Super-Resolution
- GenDR: Lighten Generative Detail Restoration
and 9 more in the interactive map.
Image restoration · 21 papers
- Analyzing the Training Dynamics of Image Restoration Transformers: A Revisit to Layer Normalization
- CHAMMI-75: pre-training multi-channel models with heterogeneous microscopy images
- Efficient Degradation-agnostic Image Restoration via Channel-Wise Functional Decomposition and Manifold Regularization
- Efficient Zero-shot Inpainting with Decoupled Diffusion Guidance
- FideDiff: Efficient Diffusion Model for High-Fidelity Image Motion Deblurring
- Learning Domain-Aware Task Prompt Representations for Multi-Domain All-in-One Image Restoration
- PI-Light: Physics-Inspired Diffusion for Full-Image Relighting
- Provably Accelerated Imaging with Restarted Inertia and Score-based Image Priors
- Reconstruct Anything Model a lightweight foundation model for computational imaging
- Turbo-DDCM: Fast and Flexible Zero-Shot Diffusion-Based Image Compression
- UniRestorer: Universal Image Restoration via Adaptively Estimating Image Degradation at Proper Granularity
- Consistency Geodesic Bridge: Image Restoration with Pretrained Diffusion Models
and 9 more in the interactive map.