The Second Challenge on Real-World Face Restoration at NTIRE 2026: Methods and Results

This project is an NTIRE (New Trends in Image Restoration and Enhancement) challenge report. While NTIRE is the gold standard for academic benchmarks in computer vision, the 'project' itself is a summary of findings rather than a proprietary software product. The defensibility is low (3) because it represents a public competition record; its value lies in the data and the comparative analysis of other people's models (like updated versions of CodeFormer or GFPGAN). The 53 forks in just 2 days indicate high researcher engagement, likely participants forking the evaluation script or baseline code. Frontier risk is high because face restoration is a core feature for smartphone OEMs (Apple, Google, Samsung) and generative AI platforms (OpenAI's Sora/DALL-E). These labs frequently absorb NTIRE winners' techniques directly into their production pipelines. The market is consolidating as these 'magic eraser' and 'photo unblur' features move from standalone apps into OS-level capabilities, leaving little room for independent startups unless they target ultra-niche forensic or medical use cases.