HiGS: History-Guided Sampling for Plug-and-Play Enhancement of Diffusion Models

Documenti analoghi

• HiWave: Training-Free High-Resolution Image Generation via Wavelet-Based Diffusion Sampling
  di: Vontobel, Tobias, et al.
  Pubblicazione: (2025)
• Eliminating Oversaturation and Artifacts of High Guidance Scales in Diffusion Models
  di: Sadat, Seyedmorteza, et al.
  Pubblicazione: (2024)
• No Training, No Problem: Rethinking Classifier-Free Guidance for Diffusion Models
  di: Sadat, Seyedmorteza, et al.
  Pubblicazione: (2024)
• Guidance in the Frequency Domain Enables High-Fidelity Sampling at Low CFG Scales
  di: Sadat, Seyedmorteza, et al.
  Pubblicazione: (2025)
• LiteVAE: Lightweight and Efficient Variational Autoencoders for Latent Diffusion Models
  di: Sadat, Seyedmorteza, et al.
  Pubblicazione: (2024)