New ICTone framework enables advanced tone style transfer with large dataset

Researchers have developed ICTone, a novel diffusion-based framework for in-context tone style transfer in photo retouching. This method addresses the limitations of existing approaches, which are hampered by a lack of large-scale, high-quality datasets with stylized ground truth. To overcome this, the team constructed TST100K, a dataset comprising 100,000 content-reference-stylized image triplets, ensuring strict stylistic consistency through a trained tone style scorer. ICTone jointly conditions on both content and reference images, leveraging semantic priors from generative models for more accurate and visually appealing color transfer. AI

IMPACT This research could lead to more sophisticated photo editing tools and improved generative models for image manipulation.

RANK_REASON The item is a research paper detailing a new method and dataset for a specific computer vision task. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• alphaXiv
• arXiv
• CatalyzeX
• CORE Recommender
• DagsHub
• Gotit.pub
• Hugging Face
• ICTone
• Influence Flower
• ScienceCast
• TST100K
• Yuhai Deng

• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →