mlsanityData to 7 Oct 2026

Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models

arXiv 2404.07724 ↗First posted 11 Apr 2024cs.CV cs.AI cs.LGcited by 240

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Guidance is a crucial technique for extracting the best performance out of image-generating diffusion models. Traditionally, a constant guidance weight has been applied throughout the sampling chain of an image. We show that guidance is clearly harmful toward the beginning of the chain (high noise levels), largely unnecessary toward the end (low noise levels), and only beneficial in the middle. We thus restrict it to a specific range of noise levels, improving both the inference speed and result quality. This limited guidance interval improves the record FID in ImageNet-512 significantly, from 1.81 to 1.40. We show that it is quantitatively and qualitatively beneficial across different sampler parameters, network architectures, and datasets, including the large-scale setting of Stable Diffusion XL. We thus suggest exposing the guidance interval as a hyperparameter in all diffusion models that use guidance.

Key facts. Cited by 240 papers in the map (data through 7 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 33 of its 42 references that are in the map. First posted on arXiv 11 Apr 2024.

Builds on

33 of 42 references.

Closest ancestors · 14

Same direction, newest first

  1. 2024Dynamical Regimes of Diffusion Models28 Feb
  2. 2023Adaptive Guidance: Training-free Acceleration of Conditional Diffusion Models19 Dec
  3. 2023Analyzing and Improving the Training Dynamics of Diffusion Models5 Dec
  4. 2023CADS: Unleashing the Diversity of Diffusion Models through Condition-Annealed Sampling26 Oct
  5. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis4 Jul
  6. 2023MDTv2: Masked Diffusion Transformer is a Strong Image Synthesizer25 Mar
  7. 2022Scalable Diffusion Models with Transformers19 Dec
  8. 2022Classifier-Free Diffusion Guidance26 Jul
  9. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  10. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  11. 2021Improved Denoising Diffusion Probabilistic Models18 Feb
  12. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
2 more
  1. 2020Denoising Diffusion Implicit Models6 Oct
  2. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 19

  1. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  2. 2023Adding Conditional Control to Text-to-Image Diffusion Modelscited by 5,466
  3. 2019Generative Modeling by Estimating Gradients of the Data Distributioncited by 4,067
  4. 2022DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generationcited by 3,286
  5. 2022DreamFusion: Text-to-3D using 2D Diffusioncited by 3,006
  6. 2022Video Diffusion Modelscited by 2,280
  7. 2022An Image is Worth One Word: Personalizing Text-to-Image Generation using Textual Inversioncited by 2,231
  8. 2022Imagen Video: High Definition Video Generation with Diffusion Modelscited by 1,765
  9. 2023Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Modelscited by 1,465
  10. 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,339
  11. 2022Magic3D: High-Resolution Text-to-3D Content Creationcited by 1,285
  12. 2019Improved Precision and Recall Metric for Assessing Generative Modelscited by 925
7 more
  1. 2022eDiff-I: Text-to-Image Diffusion Models with an Ensemble of Expert Denoiserscited by 921
  2. 2023Muse: Text-To-Image Generation via Masked Generative Transformerscited by 640
  3. 2021Grad-TTS: A Diffusion Probabilistic Model for Text-to-Speechcited by 445
  4. 20223D Neural Field Generation using Triplane Diffusioncited by 274
  5. 2023DreamBooth3D: Subject-Driven Text-to-3D Generationcited by 235
  6. 2023Exposing flaws of generative model evaluation metrics and their unfair treatment of diffusion modelscited by 199
  7. 2023Prompting Hard or Hardly Prompting: Prompt Inversion for Text-to-Image Diffusion Modelscited by 40

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2024: 31; 2025: 118; 2026: 91.

240 papers.

Sort:

By area:

  1. 2024Representation Alignment for Generation: Training Diffusion Transformers Is Easier Than You Thinkcited by 666
  2. 2025Reconstruction vs. Generation: Taming Optimization Dilemma in Latent Diffusion Modelscited by 376
  3. 2024Guiding a Diffusion Model with a Bad Version of Itselfcited by 286
  4. 2025Diffusion Transformers with Representation Autoencoderscited by 275
  5. 2024CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Modelscited by 138
  6. 2024Eliminating Oversaturation and Artifacts of High Guidance Scales in Diffusion Modelscited by 107
  7. 2025Improved Mean Flows: On the Challenges of Fastforward Generative Modelscited by 102
  8. 2025DDT: Decoupled Diffusion Transformercited by 100
  9. 2024PeRFlow: Piecewise Rectified Flow as Universal Plug-and-Play Acceleratorcited by 95
  10. 2025Representation Entanglement for Generation: Training Diffusion Transformers Is Much Easier Than You Thinkcited by 95
  11. 2025Masked Autoencoders Are Effective Tokenizers for Diffusion Modelscited by 93
  12. 2025PixelFlow: Pixel-Space Generative Models with Flowcited by 91
38 more
  1. 2025EQ-VAE: Equivariance Regularized Latent Space for Improved Generative Image Modelingcited by 85
  2. 2025Align Your Flow: Scaling Continuous-Time Flow Map Distillationcited by 81
  3. 2024Classifier-Free Guidance is a Predictor-Correctorcited by 79
  4. 2025PixNerd: Pixel Neural Field Diffusioncited by 75
  5. 2025PixelDiT: Pixel Diffusion Transformers for Image Generationcited by 75
  6. 2024Simpler Diffusion (SiD2): 1.5 FID on ImageNet512 with pixel-space diffusioncited by 74
  7. 2025Improving the Diffusability of Autoencoderscited by 74
  8. 2025Beyond Next-Token: Next-X Prediction for Autoregressive Visual Generationcited by 71
  9. 2024Analysis of Classifier-Free Guidance Weight Schedulerscited by 64
  10. 2024SoftVQ-VAE: Efficient 1-Dimensional Continuous Tokenizercited by 55
  11. 2025GigaTok: Scaling Visual Tokenizers to 3 Billion Parameters for Autoregressive Image Generationcited by 55
  12. 2025DeCo: Frequency-Decoupled Pixel Diffusion for End-to-End Image Generationcited by 55
  13. 2025No Other Representation Component Is Needed: Diffusion Transformers Can Provide Representation Guidance by Themselvescited by 50
  14. 2025CFG-Zero*: Improved Classifier-Free Guidance for Flow Matching Modelscited by 47
  15. 2025DC-AE 1.5: Accelerating Diffusion Model Convergence with Structured Latent Spacecited by 44
  16. 2025DiP: Taming Diffusion Models in Pixel Spacecited by 42
  17. 2024Mind the Time: Temporally-Controlled Multi-Event Video Generationcited by 39
  18. 2024Language-Guided Image Tokenization for Generationcited by 36
  19. 2025Flow to the Mode: Mode-Seeking Diffusion Autoencoders for State-of-the-Art Image Tokenizationcited by 35
  20. 2025Stochastic Self-Guidance for Training-Free Enhancement of Diffusion Modelscited by 35
  21. 2025UniAVGen: Unified Audio and Video Generation with Asymmetric Cross-Modal Interactionscited by 35
  22. 2024Self-Improving Diffusion Models with Synthetic Datacited by 34
  23. 2024Switti: Designing Scale-Wise Transformers for Text-to-Image Synthesiscited by 33
  24. 2025REPA Works Until It Doesn't: Early-Stopped, Holistic Alignment Supercharges Diffusion Trainingcited by 33
  25. 2026Latent Forcing: Reordering the Diffusion Trajectory for Pixel-Space Image Generationcited by 33
  26. 2024Causal Diffusion Transformers for Generative Modelingcited by 31
  27. 2024Conditional Image Synthesis with Diffusion Models: A Surveycited by 29
  28. 2025U-REPA: Aligning Diffusion U-Nets to ViTscited by 29
  29. 2025Contrastive Flow Matchingcited by 29
  30. 2024Dynamic Negative Guidance of Diffusion Modelscited by 28
  31. 2025Diffusion Models without Classifier-free Guidancecited by 28
  32. 2024SnapGen: Taming High-Resolution Text-to-Image Models for Mobile Devices with Efficient Architectures and Trainingcited by 27
  33. 2025pi-Flow: Policy-Based Few-Step Generation via Imitation Distillationcited by 25
  34. 2025SynthLight: Portrait Relighting with Diffusion Model by Learning to Re-render Synthetic Facescited by 24
  35. 2026Improved Baselines with Representation Autoencoderscited by 24
  36. 2025There is No VAE: End-to-End Pixel-Space Generative Modeling via Self-Supervised Pre-trainingcited by 23
  37. 2024Shielded Diffusion: Generating Novel and Diverse Images using Sparse Repellencycited by 21
  38. 2024Classifier-Free Guidance inside the Attraction Basin May Cause Memorizationcited by 21

50 most cited of 240 All 240 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2026On the Robustness of Distribution Support under Diffusion Guidance
  2. 2025Provable Efficiency of Guidance in Diffusion Models for General Data Distribution
  3. 2024What does guidance do? A fine-grained analysis in a simple setting
  4. 2024Theoretical Insights for Diffusion Guidance: A Case Study for Gaussian Mixture Models
  5. 2023Bridging the Gap: Addressing Discrepancies in Diffusion Model Training for Classifier-Free Guidance
  6. 2026Learned End-to-End Guidance Schedules for Diffusion Models
  7. 2026Measuring Effective Data Resolution in Guided Diffusion Posteriors
  8. 2023Selective Guidance: Are All the Denoising Steps of Guided Diffusion Important?