Related Experiment Video
Updated: Dec 1, 2025
06:49
A Standardized Approach to Extra-Oral and Intra-Oral Digital Photography
Published on: July 22, 2022
9.6K
Alignment of rendered images with photographs for testing appearance models
Applied Optics
|November 11, 2020
Summary
This study introduces a method to compare rendered images with photographs for analyzing object optical properties and appearance models. It enables quantitative validation for refining computer graphics rendering techniques.
Area of Science:
- Computer Graphics
- Material Appearance Modeling
- Computational Imaging
Background:
- Accurate analysis of physical object optical properties is crucial for realistic rendering.
- Current appearance models require robust validation against real-world data.
- Direct comparison of rendered and photographed objects is challenging due to alignment and modeling complexities.
Purpose of the Study:
- To propose a method for direct comparison of rendered images with photographs.
- To analyze the optical properties of physical objects.
- To test and refine the appropriateness of appearance models in computer graphics.
Main Methods:
- A practical method for aligning objects and light sources in rendered images to match photographs.
- Utilizing projective transformation of object edges and silhouette matching for image plane alignment.
- Introducing models for spatially varying roughness and subsurface scattering influenced by transmitted light.
Main Results:
- Successful alignment of rendered and photographed objects is demonstrated.
- Improved similarity between rendered and photographed objects through advanced appearance models.
- Quantitative validation framework for appearance models is established.
Conclusions:
- The proposed method facilitates direct, quantitative validation of appearance models.
- Advanced models for roughness and subsurface scattering enhance rendering realism.
- This work supports the progressive refinement of optical property analysis in computer graphics.
