ExpandDiff model enhances HDR reconstruction with novel diffusion technique

Researchers have developed ExpandDiff, a new diffusion model designed for single-image High Dynamic Range (HDR) reconstruction. This model addresses the challenge of inferring missing details in Low Dynamic Range (LDR) images by simultaneously reconstructing both clipped shadows and highlights. ExpandDiff utilizes a technique called Dynamic Clipping Synthesis (DCS) to create varied training data and employs a pixel-space diffusion model with spatially-adaptive normalization for prediction. In evaluations on the SI-HDR benchmark, ExpandDiff variants significantly improved HDR reconstruction accuracy, achieving up to 7.34 dB higher PU21-PSNR under two-sided clipping conditions compared to existing methods. AI

IMPACT Introduces a novel diffusion model for HDR image reconstruction, potentially improving image processing applications.

RANK_REASON The cluster contains a research paper detailing a new method for image reconstruction. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• arXiv
• Dynamic Clipping Synthesis
• ExpandDiff
• Mehmet Emre Andıran
• PU21-PSNR
• SI-HDR

• paper
• other

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