mlsanityData to 14 Oct 2026

HPSv3: Towards Wide-Spectrum Human Preference Score

arXiv 2508.03789 ↗First posted 5 Aug 2025cs.CVcited by 211

On the map: LLM alignment / RLHF / preference optimisation → Diffusion and multimodal model alignment

Abstract (arXiv)

Evaluating text-to-image generation models requires alignment with human perception, yet existing human-centric metrics are constrained by limited data coverage, suboptimal feature extraction, and inefficient loss functions. To address these challenges, we introduce Human Preference Score v3 (HPSv3). (1) We release HPDv3, the first wide-spectrum human preference dataset integrating 1.08M text-image pairs and 1.17M annotated pairwise comparisons from state-of-the-art generative models and low to high-quality real-world images. (2) We introduce a VLM-based preference model trained using an uncertainty-aware ranking loss for fine-grained ranking. Besides, we propose Chain-of-Human-Preference (CoHP), an iterative image refinement method that enhances quality without extra data, using HPSv3 to select the best image at each step. Extensive experiments demonstrate that HPSv3 serves as a robust metric for wide-spectrum image evaluation, and CoHP offers an efficient and human-aligned approach to improve image generation quality. The code and dataset are available at the HPSv3 Homepage.

Key facts. Cited by 211 papers in the map (data through 14 Oct 2026). Placed in Diffusion and multimodal model alignment, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 34 of its 43 references that are in the map. First posted on arXiv 5 Aug 2025.

Builds on

34 of 43 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025DanceGRPO: Unleashing GRPO on Visual Generation12 May
  2. 2024Deep Reward Supervisions for Tuning Text-to-Image Diffusion Models1 May
  3. 2023Human Preference Score: Better Aligning Text-to-Image Models with Human Preference25 Mar

Other references · 31

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,863
  3. 2022Hierarchical Text-Conditional Image Generation with CLIP Latentscited by 6,585
  4. 2022Photorealistic Text-to-Image Diffusion Models with Deep Language Understandingcited by 6,355
  5. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  6. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,356
  7. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  8. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,234
  9. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  10. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,614
  11. 2020Taming Transformers for High-Resolution Image Synthesiscited by 3,253
  12. 2023ImageReward: Learning and Evaluating Human Preferences for Text-to-Image Generationcited by 1,045
19 more
  1. 2024Visual Autoregressive Modeling: Scalable Image Generation via Next-Scale Predictioncited by 1,023
  2. 2023Pick-a-Pic: An Open Dataset of User Preferences for Text-to-Image Generationcited by 977
  3. 2023Human Preference Score v2: A Solid Benchmark for Evaluating Human Preferences of Text-to-Image Synthesiscited by 890
  4. 2023PixArt-α: Fast Training of Diffusion Transformer for Photorealistic Text-to-Image Synthesiscited by 864
  5. 2024PixArt-Sigma: Weak-to-Strong Training of Diffusion Transformer for 4K Text-to-Image Generationcited by 329
  6. 2022CogView2: Faster and Better Text-to-Image Generation via Hierarchical Transformerscited by 307
  7. 2024Infinity: Scaling Bitwise AutoRegressive Modeling for High-Resolution Image Synthesiscited by 304
  8. 2024Hunyuan-DiT: A Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understandingcited by 298
  9. 2024Playground v2.5: Three Insights towards Enhancing Aesthetic Quality in Text-to-Image Generationcited by 256
  10. 2023StyleGAN-T: Unlocking the Power of GANs for Fast Large-Scale Text-to-Image Synthesiscited by 212
  11. 2023JourneyDB: A Benchmark for Generative Image Understandingcited by 205
  12. 2024Fluid: Scaling Autoregressive Text-to-image Generative Models with Continuous Tokenscited by 147
  13. 2024Learning Multi-dimensional Human Preference for Text-to-Image Generationcited by 106
  14. 2021FuseDream: Training-Free Text-to-Image Generation with Improved CLIP+GAN Space Optimizationcited by 71
  15. 2022Controllable 3D Face Synthesis with Conditional Generative Occupancy Fieldscited by 37
  16. 2021Inverting Generative Adversarial Renderer for Face Reconstructioncited by 17
  17. 2025DC-AR: Efficient Masked Autoregressive Image Generation with Deep Compression Hybrid Tokenizercited by 13
  18. 2024ECNet: Effective Controllable Text-to-Image Diffusion Modelscited by 6
  19. 2024GenCA: A Text-conditioned Generative Model for Realistic and Drivable Codec Avatarscited by 3

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 49; 2026: 162.

211 papers.

Sort:

By area:

  1. 2025TempFlow-GRPO: When Timing Matters for GRPO in Flow Modelscited by 97
  2. 2025LongCat-Video Technical Reportcited by 85
  3. 2025RewardDance: Reward Scaling in Visual Generationcited by 68
  4. 2025EditScore: Unlocking Online RL for Image Editing via High-Fidelity Reward Modelingcited by 64
  5. 2025EditReward: A Human-Aligned Reward Model for Instruction-Guided Image Editingcited by 62
  6. 2025GigaWorld-0: World Models as Data Engine to Empower Embodied AIcited by 62
  7. 2025Distribution Matching Distillation Meets Reinforcement Learningcited by 50
  8. 2025Decoupled DMD: CFG Augmentation as the Spear, Distribution Matching as the Shieldcited by 45
  9. 2025StoryMem: Multi-shot Long Video Storytelling with Memorycited by 40
  10. 2026WBench: A Comprehensive Multi-turn Benchmark for Interactive Video World Model Evaluationcited by 39
  11. 2026LongCat-Next: Lexicalizing Modalities as Discrete Tokenscited by 36
  12. 2025Does FLUX Already Know How to Perform Physically Plausible Image Composition?cited by 32
38 more
  1. 2026HiDream-O1-Image: A Natively Unified Image Generative Foundation Model with Pixel-level Unified Transformercited by 32
  2. 2026SenseNova-U1: Unifying Multimodal Understanding and Generation with NEO-unify Architecturecited by 32
  3. 2025JarvisEvo: Towards a Self-Evolving Photo Editing Agent with Synergistic Editor-Evaluator Optimizationcited by 31
  4. 2025Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preferencecited by 30
  5. 2026WorldCompass: Reinforcement Learning for Long-Horizon World Modelscited by 24
  6. 2025UniPercept: Towards Unified Perceptual-Level Image Understanding across Aesthetics, Quality, Structure, and Texturecited by 22
  7. 2025Factuality Matters: When Image Generation and Editing Meet Structured Visualscited by 20
  8. 2025Multimodal RewardBench 2: Evaluating Omni Reward Models for Interleaved Text and Imagecited by 20
  9. 2025Follow-Your-Preference: Towards Preference-Aligned Image Inpaintingcited by 18
  10. 2025DiverseGRPO: Mitigating Mode Collapse in Image Generation via Diversity-Aware GRPOcited by 17
  11. 2025GARDO: Reinforcing Diffusion Models without Reward Hackingcited by 17
  12. 2026LUVE : Latent-Cascaded Ultra-High-Resolution Video Generation with Dual Frequency Expertscited by 15
  13. 2025Principled RL for Flow Matching Emerges from the Chunk-level Policy Optimizationcited by 14
  14. 2026RationalRewards: Reasoning Rewards Scale Visual Generation Both Training and Test Timecited by 14
  15. 2025Hyper-Bagel: A Unified Acceleration Framework for Multimodal Understanding and Generationcited by 13
  16. 2026Unified Personalized Reward Model for Vision Generationcited by 13
  17. 2025Consolidating Reinforcement Learning for Multimodal Discrete Diffusion Modelscited by 12
  18. 2025The Image as Its Own Reward: Reinforcement Learning with Adversarial Reward for Image Generationcited by 12
  19. 2025RealGen: Photorealistic Text-to-Image Generation via Detector-Guided Rewardscited by 12
  20. 2026Persistent Robot World Models: Stabilizing Multi-Step Rollouts via Reinforcement Learningcited by 12
  21. 2026JoyAI-Image: Awaking Spatial Intelligence in Unified Multimodal Understanding and Generationcited by 12
  22. 2026World-R1: Reinforcing 3D Constraints for Text-to-Video Generationcited by 11
  23. 2026KVPO: ODE-Native GRPO for Autoregressive Video Alignment via KV Semantic Explorationcited by 11
  24. 2025EdiVal-Agent: An Object-Centric Framework for Automated, Fine-Grained Evaluation of Multi-Turn Editingcited by 10
  25. 2026TAGRPO: Boosting GRPO on Image-to-Video Generation with Direct Trajectory Alignmentcited by 10
  26. 2026Stepwise Credit Assignment for GRPO on Flow-Matching Modelscited by 10
  27. 2025MammothModa2: A Unified AR-Diffusion Framework for Multimodal Understanding and Generationcited by 9
  28. 2025PSR: Scaling Multi-Subject Personalized Image Generation with Pairwise Subject-Consistency Rewardscited by 9
  29. 2026Leveraging Verifier-Based Reinforcement Learning in Image Editingcited by 9
  30. 2026Auto-Rubric as Reward: From Implicit Preferences to Explicit Multimodal Generative Criteriacited by 9
  31. 2025Ovis-Image Technical Reportcited by 8
  32. 2026Asymmetric Flow Modelscited by 8
  33. 2026Reinforcing Few-step Generators via Reward-Tilted Distribution Matchingcited by 8
  34. 2025ImageDoctor: Diagnosing Text-to-Image Generation via Grounded Image Reasoningcited by 7
  35. 2025DesignPref: Capturing Personal Preferences in Visual Design Generationcited by 7
  36. 2025PaCo-RL: Advancing Reinforcement Learning for Consistent Image Generation with Pairwise Reward Modelingcited by 7
  37. 2026ImagenWorld: Stress-Testing Image Generation Models with Explainable Human Evaluation on Open-ended Real-World Taskscited by 7
  38. 2026OmniNFT: Modality-wise Omni Diffusion Reinforcement for Joint Audio-Video Generationcited by 7

50 most cited of 211 All 211 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026HPSv3++: Scaling Reward Models Across the Full Spectrum of Diffusion Model Capabilities
  2. 20253DGen-Bench: Comprehensive Benchmark Suite for 3D Generative Models
  3. 2024Scalable Ranked Preference Optimization for Text-to-Image Generation
  4. 2025Text-driven 3D Human Generation via Contrastive Preference Optimization
  5. 2024SafeSora: Towards Safety Alignment of Text2Video Generation via a Human Preference Dataset
  6. 2025HuViDPO:Enhancing Video Generation through Direct Preference Optimization for Human-Centric Alignment
  7. 2024Synth-Align: Improving Trustworthiness in Vision-Language Model with Synthetic Preference Data Alignment
  8. 2024VideoDPO: Omni-Preference Alignment for Video Diffusion Generation