Photometric Stereo Systems | Active Illumination Methods

Item: Photometric Stereo Systems | Active Illumination Methods
Rating: 5
Author: PostSingularity Institute

by Shree Nayar

Formal & Physical Sciences AI — Models & Research UYQArtificial intelligence (AI)UYQVComputer vision

🎙 Shree Nayar 👥 96K 📅 April 11, 2021 ⏱ 10 min 👁 8K 📄 lecture 🧭 2026-08-17

Available in: English (current) Français

Keywords

photometric stereoactive illuminationsurface normalsreflectance estimationlight stage

Summary

This lecture by Shree Nayar from Columbia University presents photometric stereo and its extension to recover both shape and reflectance. It begins with a recap of classic photometric stereo, which uses multiple images under known lighting to compute surface normals, assuming known reflectance. The lecture then introduces the challenge of unknown reflectance parameters, requiring more measurements. A solution is demonstrated via photometric sampling, using an array of light sources and a diffuser to handle specular surfaces. The system recovers both surface normals and reflectance properties, as shown in an example with a diffuse and specular object. The lecture then discusses the Light Stage, a large-scale system for capturing human performances under rapidly varying illumination, enabling relighting and material estimation. The presenter emphasizes the use of LED lights and high-speed cameras to capture many images within a short time, allowing linear combinations for arbitrary relighting. The lecture concludes with a demonstration of relighting a person into different environments.

157 words

Critical Evaluation

Value of the Information & Strength of the Argument

The lecture provides valuable insights into advanced photometric stereo techniques, clearly explaining the principles and demonstrating practical applications. The argumentation is solid, building from basic concepts to more complex systems, with visual examples that reinforce the explanations. The presenter effectively communicates the trade-offs and innovations in the field.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high, as the content is based on established research and presented by an expert. The lecture does not cite specific sources but references known systems like the Light Stage. The title accurately reflects the content, and the presentation is well-structured and informative.

109 words

Title / Content Match

The title accurately reflects the content, which focuses on photometric stereo systems and active illumination methods.

Quality & Reliability

9/10

Lecture by a leading expert from Columbia University, presenting established methods with clear explanations and visual demonstrations. The content is consistent with academic literature and the presenter is a recognized authority in computer vision.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

First Principles of Computer Vision — Lecture series by Shree Nayar at Columbia University.

Concurring Sources

Photometric stereo - Wikipedia — General reference on photometric stereo.

Contribution & Novelties

The lecture provides a clear and concise overview of photometric stereo systems, emphasizing the extension to recover reflectance. It highlights the practical implementation of photometric sampling and the Light Stage, demonstrating their utility in relighting and material estimation. The presentation is valuable for students and practitioners seeking to understand active illumination methods.

Pour aller plus loin :

Photometric stereo - Wikipedia — Overview of the technique and its variations.
Light Stage - USC ICT — Official page of the Light Stage project.
Reflectance model - Wikipedia — General concept of reflectance and models.

93 words

Radar Profile

The radar profile shows high scores in quality and reliability, with slightly lower but still strong scores in quantity and technical level. This indicates a well-balanced, authoritative lecture that provides substantial information with clear technical depth.

Reliability 9/10