Researchers have introduced CoatFusion, a new diffusion model architecture designed for a novel image editing task called Material Coating. This task involves applying a thin layer of material onto an object in an image while preserving its geometry, distinguishing it from traditional material transfer methods. To support this, a large synthetic dataset named DataCoat110K, comprising 110,000 images with physically-based coatings, was created. Experiments demonstrate that CoatFusion outperforms existing methods in generating realistic and controllable coatings. AI
IMPACT Introduces a new image editing capability for applying realistic and controllable material coatings.
RANK_REASON Academic paper detailing a new model and dataset for image editing. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
