trends.barnas.net

method · first seen 5 Oct 2026

super-resolution

Also: FSR 1, upscaling, image upscaling, neural upsampling, AI video upscaling, blind super-resolution, resolution enhancement

Neural reconstruction of high-resolution images from low-resolution measurements.

Super-resolution reconstructs high-resolution images or video from low-resolution inputs, typically with neural networks. The sources are a scattered mix: papers on thermal, Earth-observation and semiconductor metrology super-resolution, a conformal-risk method for flagging reliable regions in scientific SR, joint SR and compression-artifact removal for industrial vision, and a Hany Farid study on whether AI upscaling helps humans read license plates. The code side is mostly gaming upscalers (FSR plugins, settings guides), a VHS restoration app built on Real-ESRGAN, and a heightmap upscaling model shown on Hacker News. No single new technique ties these together.

The field is long established. The current applied threads, calibrated uncertainty for scientific SR and human-legibility tests of AI upscaling in forensic settings, bear on whether hallucinated detail can be trusted.

Analyst note (established), 17 Nov 2026.

Mentions per week

0419 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Rank 92 on 11 Nov 2026 · 14 mentions in 28 days · 14 independent groups · papers → discussion → code · first seen 5 weeks ago · acceleration +0.67

Earliest sources

Latest sources

Mentioned together with

scanning probe microscopythermal image super-resolutionplug-and-play image restorationpartially supervised learninghomogeneous isotropic turbulencepersistence diagramtopology-informed lossground sampling distancestatic recompilationdevice-conditioned degradation modeling