FluoNeRF: Fluorescent Novel-View Synthesis Under Novel Light Source Colors and Spectra

Area of Science:

• Computer Vision
• Computer Graphics

Background:

• Synthesizing photo-realistic images from novel viewpoints and lighting is a key challenge.
• Fluorescent materials exhibit unique light absorption and emission properties, unlike reflective materials.
• Standard view synthesis methods struggle to accurately render fluorescent object colors under varied lighting.

Purpose of the Study:

• To develop a method for synthesizing photo-realistic images of scenes containing fluorescent objects.
• To enable image synthesis from novel viewpoints and under novel lighting colors and spectra.
• To overcome limitations of existing techniques in reproducing fluorescent material appearance.

Main Methods:

• Extended neural radiance fields (NeRFs) by incorporating the superposition principle of light.
• Captured sparse image sets of scenes under varying viewpoints and active lighting (color display/multi-spectral light stage).
• Synthesized images without explicit geometric or photometric modeling.

Main Results:

• Demonstrated superior performance compared to existing methods through experiments with real images.
• Confirmed the method's effectiveness in reproducing fluorescent object colors under common light sources.
• Showcased that using more than three primary colors with a light stage improves color reproduction accuracy.

Conclusions:

• The proposed method effectively synthesizes photo-realistic images of fluorescent objects under novel conditions.
• The approach advances view synthesis by accurately handling the unique spectral properties of fluorescent materials.
• This work provides a robust solution for realistic rendering of fluorescent objects in computer graphics and vision.