kwonminki/One-sentence_Diffusion_summary

Provides curated summaries of recent diffusion model papers with technical implementation details—covering text-to-image generation, image editing via inversion and exemplar guidance, video synthesis, and architectural innovations like ControlNet for conditional control. Organized chronologically with Korean explanations of key mechanisms (e.g., CLIP space direction manipulation, cross-attention layer-specific prompting, temporal shift blocks for video). Targets the computer vision community studying diffusion-based generation and editing workflows.

429 stars. No commits in the last 6 months.