Bigger is not Always Better: Scaling Properties of Latent Diffusion Models

Documenti analoghi

• CoDi: Conditional Diffusion Distillation for Higher-Fidelity and Faster Image Generation
  di: Mei, Kangfu, et al.
  Pubblicazione: (2023)
• The Power of Context: How Multimodality Improves Image Super-Resolution
  di: Mei, Kangfu, et al.
  Pubblicazione: (2025)
• Inversion by Direct Iteration: An Alternative to Denoising Diffusion for Image Restoration
  di: Delbracio, Mauricio, et al.
  Pubblicazione: (2023)
• Denoising: A Powerful Building-Block for Imaging, Inverse Problems, and Machine Learning
  di: Milanfar, Peyman, et al.
  Pubblicazione: (2024)
• The Geometry of Noise: Why Diffusion Models Don't Need Noise Conditioning
  di: Sahraee-Ardakan, Mojtaba, et al.
  Pubblicazione: (2026)