Real-time computer-generated integral imaging light field displays: revisiting the point retracing rendering method...

Area of Science:

• Computer graphics
• 3D display technology
• Computational imaging

Background:

• Integral imaging light field displays (InIm-LFDs) create realistic 3D images using elemental image arrays (EIAs).
• Real-time EIA generation is computationally intensive, hindering adoption with entry-level hardware.
• Existing viewpoint-based rendering methods face challenges in speed and complexity.

Purpose of the Study:

• To develop a computationally efficient method for real-time EIA generation for InIm-LFDs.
• To reduce the computational complexity of 3D rendering for computer-generated integral imaging.
• To overcome the tradeoff between rendering speed, accuracy, and system complexity.

Main Methods:

• Revisiting point retracing rendering and treating InIm-LFDs and 2D displays as sharing sampling and reconstruction phases.
• Utilizing static voxels for signal sampling and homogeneous pixels for signal reconstruction.
• Employing a pre-computed voxel-pixel mapping lookup table (LUT) derived from raytracing.

Main Results:

• The proposed method resamples 3D data with voxels and uses the LUT to assign values to homogeneous pixels.
• Computational complexity is reduced by several orders of magnitude.
• Experimental rendering speeds of 7-10 ms per full-HD EIA frame achieved on an entry-level laptop.
• Fourier domain analysis confirms negligible differences in output signal frequency spectrum compared to viewpoint-based methods, even with sampling errors.

Conclusions:

• The proposed voxel-based rendering method offers a significant speedup for EIA generation.
• It effectively breaks the conventional tradeoff between rendering speed, accuracy, and system complexity in computer-generated integral imaging.
• The method shows promise for enabling realistic 3D displays on accessible computing hardware.