My Research Journey at CIPLAB

Jinwoo Kim - CIP Lab, CS, Yonsei University

This talk provides a first-principles introduction to camera systems, motivated by modern research problems such as image super-resolution, video restoration, and diffusion-based imaging pipelines. Starting from the fundamental question of how images are formed, we revisit film cameras to understand the physical foundations of focus and exposure, including lens movement, depth of field, and the exposure triangle (ISO, shutter speed, and aperture). We then transition to digital cameras, where these principles are extended through computational mechanisms such as autofocus (phase detection and contrast-based methods) and metering strategies. By unifying the concepts of “recognition” and “control” in autofocus, and linking exposure to perceptual assumptions like 18% gray, the talk builds an intuitive yet rigorous understanding of how cameras capture images. Ultimately, this presentation aims to bridge classical camera principles with modern computational imaging and machine learning, highlighting why a deep understanding of camera systems is essential for designing reliable and faithful image processing models.

PPT CV

Catering Courtesy of CIP Lab