Research map / Video / Video segmentation and tracking
Pix2HDR - A Pixel-Wise Acquisition and Deep Learning-Based Synthesis Approach for High-Speed HDR Videos
Caixin Wang, Jie J. Zhang, Matthew A. Wilson, Ralph Etienne‐Cummings
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace spatial box filter synthesis with LDR-HDR 3D CNN networks (ablation) | MPVE sampling · HDR-VDP/HDR-VQM comparison; rotating H | temporal contrast: 0.06 -> 0.2 | Improved |
| replace box filter with learned LDR-HDR synthesis (also trained at 1:4) (ablation) | Box filter (1:4) vs Pix2HDR (1:4) · HDM-HDR-2014 (Poker Fullshot test) | PSNR (dB): 32.99 -> 43.51 | Improved |
| replace interleaved long-short exposure sampling with MPVE sampling | Chen21 (1:16) vs Pix2HDR (1:16) · HDM-HDR-2014 (Poker Fullshot test) | HDR-VQM: 54.61 -> 66.95 | Improved |
| replace interleaved long-short exposure sampling with MPVE sampling | Chen21 (1:16) vs Pix2HDR (1:16) · HDM-HDR-2014 (Poker Fullshot test) | SSIM: 0.9364 -> 0.9782 | Improved |
| reduce min-max exposure ratio from 1:16 to 1:4 | Pix2HDR · HDM-HDR-2014 (Poker Fullshot test) | PSNR (dB): 42.66 -> 43.51 | Improved |
| reduce min-max exposure ratio from 1:8 to 1:4 | Pix2HDR · HDM-HDR-2014 (Poker Fullshot test) | HDR-VQM: 67.62 -> 68.96 | Improved |
| increase exposure ratio from 1:4 to 1:16 for dynamic range | MPVE sampling · ISO-21550 target | dynamic range (dB): 36 dB -> 60 dB | Improved |
| add phase-offset multi-pixel sampling (MPVE) over single pixel timing | PE-CMOS pixels · LED two-event test | temporal resolution: 4 ms -> 2 ms | Improved |
| combine short/medium/long exposures via Pix2HDR instead of single long exposure | single-exposure subframes · low light rotating H | temporal contrast: 0.246 -> 0.307 | Improved |
| combine exposures via Pix2HDR instead of short exposure subframe | single-exposure subframes · bright background rotating H | temporal contrast: 0.493 -> 0.531 | Improved |
| use long exposure under bright background | long-exposure subframes · bright background rotating H | temporal contrast: 0.280 -> 0.213 vs 0.493 and 0.485 for short/medium | Hurt |
| use long exposure instead of short in low light | single-exposure subframes · low light rotating H | full duration (ms): 172 -> 190 | Hurt |
Closest papers by meaning
- HDR Video Reconstruction: A Coarse-To-Fine Network and a Real-World Benchmark Dataset (ICCV 2021 · similarity 0.91)
- HDRFlow: Real-Time HDR Video Reconstruction with Large Motions (CVPR 2024 · similarity 0.89)
- Towards Real-World HDR Video Reconstruction: A Large-Scale Benchmark Dataset and A Two-Stage Alignment Network (CVPR 2024 · similarity 0.89)
- Single Shot HDR Recovery via a Video Diffusion Prior (NeurIPS 2026 · similarity 0.89)
- High Dynamic Range Novel View Synthesis with Single Exposure (ICML 2025 · similarity 0.88)
- HDR-GS: Efficient High Dynamic Range Novel View Synthesis at 1000x Speed via Gaussian Splatting (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
