Zero-Order Diffraction Suppression in Full Field-of-View Computer Generated Holography: A Camera In the Loop Interferometric Approach

TL;DR

This work tackles zero-order diffraction (ZOD) artifacts in full-field, phase-only CGH for AR/NED applications. It introduces a camera-in-the-loop (CITL) interferometric approach that creates a ZOD replica beam in a plane conjugated to the SLM and uses CITL to recover a pixel-wise corrective phase map that cancels ZOD through destructive interference, while preserving the SLM's full modulation depth. Experiments on point-cloud and 2D/3D holograms demonstrate up to $99\%$ ZOD suppression with minimal impact on image quality and field of view, and the calibration generalizes across holograms for real-time operation. The method eliminates bulky filtering optics and enables compact, high-fidelity holographic engines for AR/MR displays, with potential extensions to RGB color and aberration compensation in diffractive optics.

Abstract

We introduce a novel interferometric approach for suppressing zero-order diffraction (ZOD) in phase-only computer-generated holography. The technique relies on the destructive interference between the zeroth-order light and a suppression beam in a plane optically conjugated to the spatial light modulator (SLM). A camera-in-the-loop (CITL) calibration procedure retrieves the optimal pixel-wise phase map that cancels out the ZOD component with high precision, while preserving the full modulation depth of the SLM. Experimental demonstrations on point-cloud and 2D/3D holograms achieve up to 99% suppression of the ZOD intensity, without loss of image quality or field of view. Once calibrated, the correction can be applied to any hologram without recomputation, enabling real-time operation and robust performance over time. This method removes a long-standing barrier to the practical deployment of full-field holography, facilitating the development of compact, high-fidelity holographic engines for augmented and mixed reality displays.

Figures (3)

• Figure 1: (a) Schematic of the 4f setup with unitary magnification. A 300 $\mu$m pinhole placed at the Fourier plane acts as a low-pass filter, allowing only the zero-order diffraction (ZOD) and the corrective beam to pass. The camera records the spatial interference between these two beams. (b) The recorded intensity for three representative pixel follow the expected cosinusoidal dependence on the applied piston phase. The minima occur for different values of the phase shift, therefore, allowing for an optimal bidimensional corrective phase map. (c) The integral of the pixel intensity over the entire camera sensor in the 4f configuration decreases with each iteration and quickly converges, with negligible improvement after only a few cycles. The retrieved phase maps at successive iterations are also shown, highlighting that the updates become progressively smaller. (d) Schematic representation of the employed algorithm for ideal phase retrivial.
• Figure 2: Experimental demonstration of ZOD suppression. (a) Reconstructed point-cloud hologram without correction and (b) with correction. The magnified view of the ZOD central region in (a) shows a bright ZOD peak, while the corresponding view in (b) shows its elimination. (c) Quantitative intensity profile extracted from the magnified insets of (a) and (b), comparing the uncorrected (blue dashed line) and corrected (orange dashed line) cases. (d) Reconstructed tree hologram without correction and (e) with correction. The magnified box in (d) shows the ZOD, which is removed in (e). For visibility reasons ZOD of the uncorrected case is intentionally allowed to saturate: both figures are normalized to the maximum value of the uncorrected holographic image (d), excluding the ZOD region. (f) Quantitative intensity comparison for the tree hologram's central region for non saturating image. Image quality metrics (PV uniformity, PSNR, SSIM, MSE) are nearly identical for the uncorrected and corrected holograms. These results confirm that the proposed correction consistently suppresses the ZOD without degrading the holographic image quality.
• Figure 3: Qualitative assessment of ZOD suppression in multi-plane holographic projections within a mixed-reality environment. Panels (a)–(c) illustrate uncorrected holograms exhibiting haze and reduced contrast across focal planes: (a) closest to the observer, (b) intermediate distance, and (c) farthest from the observer. Panels (d)–(f) show corresponding corrected projections with ZOD suppression, yielding sharper, high-contrast images without central artifacts. Insets provide magnified views of ZOD region highlighting enhanced fidelity. The setup confirms robust 3D rendering for augmented reality, with ZOD elimination essential for artifact-free multi-depth visualization.