SIGGRAPH 2025 Paper: Turning Motion-Blurred Photos into Coherent Videos
jon_barron · x · 2026-08-11
Blur2Vid is a research paper accepted to SIGGRAPH Asia 2025. By post-training a video model, the method extracts motion information from motion-blurred photos (caused by hand shake or movement) and converts it into a coherent video sequence, effectively bringing historical blurry photos to life.
Related event: Blur2Vid Model Converts Blurry Photos into Coherent Videos(2 posts)→
More from Multimodal
- Runway Integrates Ideogram for 0.6-Second AI Image Generation — runwayml · 2026-08-11
- Security Researcher Jeremi Gosney Says Goodbye to Suno AI — nptacek · 2026-08-11
- ACE-Step Workflow: Negative Prompts Are Key to Predictable Background Music — Illustrious_Usual_10 · 2026-08-11
- AI Video Generation Tip: Sync with Audio Using Reference Images and Prompt — techhalla · 2026-08-11
- Agent Workflow Generates 2K Images and Multi-Angle Character Sheets in Minutes — techhalla · 2026-08-11
- Workflow Sharing: Creating AI Music Videos with the H3 Model — techhalla · 2026-08-11