mlsanityData to 14 Oct 2026

Dynamic Negative Guidance of Diffusion Models

arXiv 2410.14398 ↗First posted 18 Oct 2024cs.CVcited by 28

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Negative Prompting (NP) is widely utilized in diffusion models, particularly in text-to-image applications, to prevent the generation of undesired features. In this paper, we show that conventional NP is limited by the assumption of a constant guidance scale, which may lead to highly suboptimal results, or even complete failure, due to the non-stationarity and state-dependence of the reverse process. Based on this analysis, we derive a principled technique called Dynamic Negative Guidance, which relies on a near-optimal time and state dependent modulation of the guidance without requiring additional training. Unlike NP, negative guidance requires estimating the posterior class probability during the denoising process, which is achieved with limited additional computational overhead by tracking the discrete Markov Chain during the generative process. We evaluate the performance of DNG class-removal on MNIST and CIFAR10, where we show that DNG leads to higher safety, preservation of class balance and image quality when compared with baseline methods. Furthermore, we show that it is possible to use DNG with Stable Diffusion to obtain more accurate and less invasive guidance than NP.

Key facts. Cited by 28 papers in the map (data through 14 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 40 of its 50 references that are in the map. First posted on arXiv 18 Oct 2024.

Builds on

40 of 50 references.

Closest ancestors · 19

Same direction, newest first

  1. 2024Analysis of Classifier-Free Guidance Weight Schedulers19 Apr
  2. 2024Applying Guidance in a Limited Interval Improves Sample and Distribution Quality in Diffusion Models11 Apr
  3. 2024Towards Memorization-Free Diffusion Models1 Apr
  4. 2024Understanding and Improving Training-free Loss-based Diffusion Guidance19 Mar
  5. 2024Critical windows: non-asymptotic theory for feature emergence in diffusion models3 Mar
  6. 2024Dynamical Regimes of Diffusion Models28 Feb
  7. 2024A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Data26 Feb
  8. 2023Adaptive Guidance: Training-free Acceleration of Conditional Diffusion Models19 Dec
  9. 2023The statistical thermodynamics of generative diffusion models: Phase transitions, symmetry breaking and critical instability26 Oct
  10. 2023In search of dispersed memories: Generative diffusion models are associative memory networks29 Sep
  11. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis4 Jul
  12. 2023Spontaneous Symmetry Breaking in Generative Diffusion Models31 May
7 more
  1. 2023FreeDoM: Training-Free Energy-Guided Conditional Diffusion Model17 Mar
  2. 2023Reduce, Reuse, Recycle: Compositional Generation with Energy-Based Diffusion Models and MCMC22 Feb
  3. 2022Classifier-Free Diffusion Guidance26 Jul
  4. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  5. 2021Diffusion Models Beat GANs on Image Synthesis11 May
  6. 2021Maximum Likelihood Training of Score-Based Diffusion Models22 Jan
  7. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 21

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  3. 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,152
  4. 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,354
  5. 2023Erasing Concepts from Diffusion Modelscited by 539
  6. 2022Safe Latent Diffusion: Mitigating Inappropriate Degeneration in Diffusion Modelscited by 513
  7. 2024Lumiere: A Space-Time Diffusion Model for Video Generationcited by 455
  8. 2023Unified Concept Editing in Diffusion Modelscited by 388
  9. 2020Visual Transformers: Token-based Image Representation and Processing for Computer Visioncited by 334
  10. 2023Ablating Concepts in Text-to-Image Diffusion Modelscited by 317
  11. 2023Your Diffusion Model is Secretly a Zero-Shot Classifiercited by 306
  12. 2023Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Modelscited by 301
9 more
  1. 2017GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibriumcited by 213
  2. 2023Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Modelscited by 177
  3. 2023Re-imagine the Negative Prompt Algorithm: Transform 2D Diffusion into 3D, alleviate Janus problem and Beyondcited by 139
  4. 2021VALSE: A Task-Independent Benchmark for Vision and Language Models Centered on Linguistic Phenomenacited by 135
  5. 2023Circumventing Concept Erasure Methods For Text-to-Image Generative Modelscited by 85
  6. 2024Get What You Want, Not What You Don't: Image Content Suppression for Text-to-Image Diffusion Modelscited by 63
  7. 2024Understanding the Impact of Negative Prompts: When and How Do They Take Effect?cited by 51
  8. 2024Learn "No" to Say "Yes" Better: Improving Vision-Language Models via Negationscited by 46
  9. 2024How and where does CLIP process negation?cited by 16

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 2; 2025: 15; 2026: 11.

28 papers.

Sort:

By area:

  1. 2025Contrastive Flow Matchingcited by 29
  2. 2025Training-Free Safe Denoisers for Safe Use of Diffusion Modelscited by 21
  3. 2025Composition and Control with Distilled Energy Diffusion Models and Sequential Monte Carlocited by 20
  4. 2025Visual Generation Without Guidancecited by 17
  5. 2025Feedback Guidance of Diffusion Modelscited by 16
  6. 2025Learn to Guide Your Diffusion Modelcited by 14
  7. 2026When Safety Collides: Resolving Multi-Category Harmful Conflicts in Text-to-Image Diffusion via Adaptive Safety Guidancecited by 11
  8. 2025A Comprehensive Survey on Concept Erasure in Text-to-Image Diffusion Modelscited by 9
  9. 2026The Entropic Signature of Class Speciation in Diffusion Modelscited by 9
  10. 2026Multimodal Unlearning Across Vision, Language, Video, and Audio: Survey of Methods, Datasets, and Benchmarkscited by 7
  11. 2025Measuring Semantic Information Production in Generative Diffusion Modelscited by 6
  12. 2026Breaking Lock-In: Preserving Steerability under Low-Data VLA Post-Trainingcited by 6
16 more
  1. 2025VSF: Simple, Efficient, and Effective Negative Guidance in Few-Step Image Generation Models By Value Sign Flipcited by 4
  2. 2025Selective Classifier-free Guidance for Zero-shot Text-to-speechcited by 4
  3. 2026Emergence of Distortions in High-Dimensional Guided Diffusion Modelscited by 4
  4. 2025Synthetic Data Augmentation using Pre-trained Diffusion Models for Long-tailed Food Image Classificationcited by 3
  5. 2025NegFaceDiff: The Power of Negative Context in Identity-Conditioned Diffusion for Synthetic Face Generationcited by 3
  6. 2026Failing Forward: Adaptive Failure-Informed Learning for Vision-Language-Action Modelscited by 3
  7. 2024SafeCFG: Controlling Harmful Features with Dynamic Safe Guidance for Safe Generationcited by 2
  8. 2025Not Every Time and Frequency Need to Be Forgotten in Diffusion Unlearningcited by 2
  9. 2026EGLOCE: Training-Free Energy-Guided Latent Optimization for Concept Erasurecited by 1
  10. 2024ContrastiveCFG: Guiding Diffusion Sampling by Contrasting Positive and Negative Conceptscited by 0
  11. 2025Dynamic VLM-Guided Negative Prompting for Diffusion Modelscited by 0
  12. 2025Semantic Anchoring for Robust Personalization in Text-to-Image Diffusion Modelscited by 0
  13. 2026Concept Removal Guidance: Evidence-Calibrated Negative Guidance for Safe Diffusion Samplingcited by 0
  14. 2026Optimal Self-Distillation for Rectified Flow via Linear Probingcited by 0
  15. 2026Signed Rectified Flow: Negativity-Controlled Generationcited by 0
  16. 2026Specificity-Aware Diffusion Steering via Variance-Reduced Sequential Monte Carlocited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Diffusion Models with Adaptive Negative Sampling Without External Resources
  2. 2025Normalized Attention Guidance: Universal Negative Guidance for Diffusion Models
  3. 2024Supercharged One-step Text-to-Image Diffusion Models with Negative Prompts
  4. 2025Entropy Rectifying Guidance for Diffusion and Flow Models
  5. 2023Null-text Guidance in Diffusion Models is Secretly a Cartoon-style Creator
  6. 2022Improving Sample Quality of Diffusion Models Using Self-Attention Guidance
  7. 2025Dynamic Classifier-Free Diffusion Guidance via Online Feedback
  8. 2025REG: Rectified Gradient Guidance for Conditional Diffusion Models