Reproduce “HIVE-Diff: High Frame Rate HDR Video Reconstruction from a Single LDR Frame and Events” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Caixin Wang, Qingrui Wang, Jie Zhang, Ralph Etienne-Cummings
HIVE-Diff reconstructs multiple HDR video frames from one low-dynamic-range reference and events. It uses events to guide temporal generation and HDR completion, with a pretrained video prior and real-domain fine-tuning on LDR-event sequences. On five held-out EventHDR sequences, its five-frame model achieves PSNR 17.98 and temporal warping error 0.0065; clipped content without events is prior-generated, not measured.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.