06/11/2025, 03:00 PM UTC
画作受损？用AI生成的“蒙版”几小时即可修复Have a damaged painting? Restore it in just hours with an AI-generated “mask”
➀ 麻省理工学院（MIT）研究人员Alex Kachkine开发了一种AI驱动的修复方法，通过打印可移除的聚合物“ 蒙版”精准匹配颜色，将修复数千处损伤的时间缩短至3.5小时；
➁ 该方法结合数字扫描、AI分析和双层打印蒙版（颜色层+白色层），可对齐原始画作并保留数字记录供未来修复参考；
➂ 强调需与艺术修复师合作以确保还原符合艺术家原意，该方法有望让更多受损藏品重见天日。
➀ MIT researcher Alex Kachkine developed an AI-powered method to restore damaged paintings by printing a removable polymer mask with precise color matching, reducing restoration time to 3.5 hours for thousands of damaged areas;
➁ The method combines digital scans, AI analysis, and a two-layer printable mask (color + white) to align with original artworks, preserving a digital record for future conservators;
➂ Ethical considerations emphasize collaboration with art conservators to ensure restorations respect the artist’s intent, potentially revitalizing stored damaged artworks for public display.
---
本文由大语言模型（LLM）生成，旨在为读者提供半导体新闻内容的知识扩展（Beta）。
