Perspective on the development and application of light-field cameras in flow diagnostics

Zu Puayen Tan*, Brian S. Thurow

*Corresponding author for this work

Research output: Contribution to journalReview articlepeer-review

6 Scopus citations


Multi-camera flow diagnostics have made large gains in recent years in the field of three-dimensional and multi-physics measurements. However, cost, complexity and optical access pose challenges that place multi-camera techniques out of reach for many labs. In that context, light-field (LF) imaging represents an alternative approach that can potentially alleviate some of these challenges. LF flow diagnostics is a branch of measurement techniques introduced within the last decade that are based on a plenoptic camera's unique ability to capture three-dimensional and multi-spectral data via a single objective lens and image sensor-albeit at reduced lateral resolutions and depth-to-lateral uncertainty ratios of 3-10 due to limited parallax angle. Thus far, LF flow diagnostics have successfully achieved significant camera-reduction alongside other performance improvements in 3D flow velocimetry, 3D particle tracking, 3D scalar-field tomography, micro-fluidic velocimetry and multi-spectral imaging, as well as early demonstrations of single-camera multi-physics measurements for applications such as 3D fluid-structure interactions. Here, we discuss the state of development in LF flow diagnostics, highlight on-going challenges, and project potential advancements in the near future.

Original languageEnglish
Article number101001
Pages (from-to)1-24
Number of pages24
JournalMeasurement Science and Technology
Issue number10
StatePublished - Oct 2021


  • 3D background-oriented schlieren
  • 3D flow diagnostics
  • light-field
  • multi-spectral imaging
  • particle image velocimetry
  • plenoptic
  • scalar-field tomography


Dive into the research topics of 'Perspective on the development and application of light-field cameras in flow diagnostics'. Together they form a unique fingerprint.

Cite this