mlsanityData to 7 Oct 2026

SPG: Improving Motion Diffusion by Smooth Perturbation Guidance

arXiv 2503.02577 ↗First posted 4 Mar 2025cs.CVcited by 2

On the map: Diffusion models → Video generation

Abstract (arXiv)

This paper presents a test-time guidance method to improve the output quality of the human motion diffusion models without requiring additional training. To have negative guidance, Smooth Perturbation Guidance (SPG) builds a weak model by temporally smoothing the motion in the denoising steps. Compared to model-agnostic methods originating from the image generation field, SPG effectively mitigates out-of-distribution issues when perturbing motion diffusion models. In SPG guidance, the nature of motion structure remains intact. This work conducts a comprehensive analysis across distinct model architectures and tasks. Despite its extremely simple implementation and no need for additional training requirements, SPG consistently enhances motion fidelity. Project page can be found at https://spg-blind.vercel.app/

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Placed in Video generation, one of 14 directions in Diffusion models. Builds on 44 of its 52 references that are in the map. First posted on arXiv 4 Mar 2025.

Builds on

44 of 52 references.

Closest ancestors · 14

Same direction, newest first

  1. 2024Flexible Motion In-betweening with Diffusion Models17 May
  2. 2023AAMDM: Accelerated Auto-regressive Motion Diffusion Model2 Dec
  3. 2023Stable Video Diffusion: Scaling Latent Video Diffusion Models to Large Datasets25 Nov
  4. 2023DynamiCrafter: Animating Open-domain Images with Video Diffusion Priors18 Oct
  5. 2023Enhanced Fine-grained Motion Diffusion for Text-driven Human Motion Synthesis23 May
  6. 2023Guided Motion Diffusion for Controllable Human Motion Synthesis21 May
  7. 2023Align your Latents: High-Resolution Video Synthesis with Latent Diffusion Models18 Apr
  8. 2023ReMoDiffuse: Retrieval-Augmented Motion Diffusion Model3 Apr
  9. 2023Human Motion Diffusion as a Generative Prior2 Mar
  10. 2023Single Motion Diffusion12 Feb
  11. 2023Modiff: Action-Conditioned 3D Motion Generation with Denoising Diffusion Probabilistic Models10 Jan
  12. 2022Executing your Commands via Motion Diffusion in Latent Space8 Dec
2 more
  1. 2022Human Motion Diffusion Model29 Sep
  2. 2022MotionDiffuse: Text-Driven Human Motion Generation with Diffusion Model31 Aug

Other references · 30

  1. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  2. 2021Diffusion Models Beat GANs on Image Synthesiscited by 8,185
  3. 2022Scalable Diffusion Models with Transformerscited by 5,931
  4. 2022Classifier-Free Diffusion Guidancecited by 5,717
  5. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesiscited by 4,290
  6. 2023MotionGPT: Human Motion as a Foreign Languagecited by 521
  7. 2021Action-Conditioned 3D Human Motion Synthesis with Transformer VAEcited by 497
  8. 2020Action2Motion: Conditioned Generation of 3D Human Motionscited by 438
  9. 2016The KIT Motion-Language Datasetcited by 334
  10. 2024Guiding a Diffusion Model with a Bad Version of Itselfcited by 286
  11. 2023OmniControl: Control Any Joint at Any Time for Human Motion Generationcited by 233
  12. 2022MoFusion: A Framework for Denoising-Diffusion-based Motion Synthesiscited by 218
18 more
  1. 2022Listen, Denoise, Action! Audio-Driven Motion Synthesis with Diffusion Modelscited by 188
  2. 2024InstantStyle: Free Lunch towards Style-Preserving in Text-to-Image Generationcited by 166
  3. 2022Improving Sample Quality of Diffusion Models Using Self-Attention Guidancecited by 151
  4. 2024Implicit Style-Content Separation using B-LoRAcited by 135
  5. 2024Self-Rectifying Diffusion Sampling with Perturbed-Attention Guidancecited by 108
  6. 2023Avatars Grow Legs: Generating Smooth Human Motion from Sparse Tracking Inputs with Diffusion Modelcited by 104
  7. 2024Visual Style Prompting with Swapping Self-Attentioncited by 62
  8. 2022MoDi: Unconditional Motion Synthesis from Diverse Datacited by 59
  9. 2024Smoothed Energy Guidance: Guiding Diffusion Models with Reduced Energy Curvature of Attentioncited by 55
  10. 2021Rhythm is a Dancer: Music-Driven Motion Synthesis with Global Structurecited by 53
  11. 2024No Training, No Problem: Rethinking Classifier-Free Guidance for Diffusion Modelscited by 52
  12. 2024RoHM: Robust Human Motion Reconstruction via Diffusioncited by 46
  13. 2024Spatiotemporal Skip Guidance for Enhanced Video Diffusion Samplingcited by 33
  14. 2023Plan, Posture and Go: Towards Open-World Text-to-Motion Generationcited by 26
  15. 2024Self-Guidance: Boosting Flow and Diffusion Generation on Their Owncited by 22
  16. 2019A Large-scale Varying-view RGB-D Action Dataset for Arbitrary-view Human Action Recognitioncited by 16
  17. 2024SCENIC: Scene-aware Semantic Navigation with Instruction-guided Controlcited by 13
  18. 2024Constrained Diffusion with Trust Samplingcited by 9

Built on by

Cited by year

Citing papers per year: 2025: 2.

2 papers.

  1. 2025Stochastic Self-Guidance for Training-Free Enhancement of Diffusion Modelscited by 35
  2. 2025AudioMoG: Guiding Audio Generation with Mixture-of-Guidancecited by 5

Similar, not cited

Close in wording, no citation either way

  1. 2023Optimizing Diffusion Noise Can Serve As Universal Motion Priors
  2. 2024Motion Guidance: Diffusion-Based Image Editing with Differentiable Motion Estimators
  3. 2022PhysDiff: Physics-Guided Human Motion Diffusion Model
  4. 2025Taming Video Models for 3D and 4D Generation via Zero-Shot Camera Control
  5. 2024Guiding a diffusion model using sliding windows
  6. 2025Taming Video Diffusion Prior with Scene-Grounding Guidance for 3D Gaussian Splatting from Sparse Inputs
  7. 2026Correcting Guided Diffusion Trajectories with Spectral Alignment
  8. 2024MotionShop: Zero-Shot Motion Transfer in Video Diffusion Models with Mixture of Score Guidance