Where and How to Perturb: On the Design of Perturbation Guidance in Diffusion and Flow Models

Documenti analoghi

• A Noise is Worth Diffusion Guidance
  di: Ahn, Donghoon, et al.
  Pubblicazione: (2024)
• Self-Rectifying Diffusion Sampling with Perturbed-Attention Guidance
  di: Ahn, Donghoon, et al.
  Pubblicazione: (2024)
• TAG: Tangential Amplifying Guidance for Hallucination-Resistant Sampling
  di: Cho, Hyunmin, et al.
  Pubblicazione: (2025)
• Retrieval-Augmented Score Distillation for Text-to-3D Generation
  di: Seo, Junyoung, et al.
  Pubblicazione: (2024)
• APPLE: Attribute-Preserving Pseudo-Labeling for Diffusion-Based Face Swapping
  di: Kang, Jiwon, et al.
  Pubblicazione: (2026)