Light Transport Simulation Research Articles

Deep learning-driven optimization of real-time ray tracing for enhanced immersive virtual reality

Photorealistic rendering is essential for immersive virtual reality and real-time ray tracing is required for this, yet the necessary computational load has so far hindered its use in scenarios where a fast response and high frame rates are paramount. To overcome this challenge, we present a deep learning-based model that optimises the computational load of ray tracing through utilising convolutional neural networks (CNNs) to predict lightscene interactions. We achieve this by training CNNs from datasets with offline ray-traced images. Using this learning process, we approximate the contents of light transport simulations from scene point sampling. The use of this optimised algorithm in virtual reality scenes, thus, enables renderings of more complex scenes, with dynamic lighting, complex geometry and interactive elements under high frame rates, keeping users immersed and responsive to events. Our results show that this method achieves visual similarities to traditional ray tracing, which is photorealistic rendering, but can be performed in real-time in virtual reality. We anticipate this work leads to many potential applications of ray tracing in many scenarios in virtual reality, such as gaming, architectural rendering and training.

Target-Aware Image Denoising for Inverse Monte Carlo Rendering

Physically based differentiable rendering allows an accurate light transport simulation to be differentiated with respect to the rendering input, i.e., scene parameters, and it enables inferring scene parameters from target images, e.g., photos or synthetic images, via an iterative optimization. However, this inverse Monte Carlo rendering inherits the fundamental problem of the Monte Carlo integration, i.e., noise, resulting in a slow optimization convergence. An appealing approach to addressing such noise is exploiting an image denoiser to improve optimization convergence. Unfortunately, the direct adoption of existing image denoisers designed for ordinary rendering scenarios can drive the optimization into undesirable local minima due to denoising bias. It motivates us to reformulate a new image denoiser specialized for inverse rendering. Unlike existing image denoisers, we conduct our denoising by considering the target images, i.e., specific information in inverse rendering. For our target-aware denoising, we determine our denoising weights via a linear regression technique using the target. We demonstrate that our denoiser enables inverse rendering optimization to infer scene parameters robustly through a diverse set of tests.

A Fully-correlated Anisotropic Micrograin BSDF Model

We introduce an improved version of the micrograin BSDF model [Lucas et al. 2023] for the rendering of anisotropic porous layers. Our approach leverages the properties of micrograins to take into account the correlation between their height and normal, as well as the correlation between the light and view directions. This allows us to derive an exact analytical expression for the Geometrical Attenuation Factor (GAF), summarizing shadowing and masking inside the porous layer. This fully-correlated GAF is then used to define appropriate mixing weights to blend the BSDFs of the porous and base layers. Furthermore, by generalizing the micrograins shape to anisotropy, combined with their fully-correlated GAF, our improved BSDF model produces effects specific to porous layers such as retro-reflection visible on dust layers at grazing angles or height and color correlation that can be found on rusty materials. Finally, we demonstrate very close matches between our BSDF model and light transport simulations realized with explicit instances of micrograins, thus validating our model.

On the Origin of the Photoplethysmography Signal: Modeling of Volumetric and Aggregation Effects

This study aimed to examine the mechanisms of the photoplethysmography (PPG) signal formation using Monte Carlo simulations of light transport in biological tissues and experimental observations. Based on a three-layer skin model in backscattering geometry, we sequentially simulated volumetric blood changes and the aggregation/disaggregation of erythrocytes in the dermal layer and estimated their contribution to the registered PPG signal. The calculations were conducted for two wavelengths: 525 nm and 810 nm. For green light, absorption predominates over scattering in the formation of a PPG signal, whereas, for near-infrared light, scattering prevails over absorption. This theoretical result was verified using the Modified Beer–Lambert law and clinical in vivo PPG data of seven healthy subjects. Changes in the size of the scatterers during erythrocyte aggregation and disaggregation can significantly contribute to the PPG signal at near-infrared light. Thus, for the green waveband, the classical volumetric model can be considered dominant in the PPG signal formation. In contrast, for the near-infrared range, both volumetric and aggregation effects must be considered as being approximately equal.

Monte Carlo simulations of light transport in sunscreen formulations

Sunscreens are used for the protection of human skin against the harmful effects of solar UV radiation. Due to the low thickness of sunscreen films typically applied to the skin, it can be challenging to achieve the strong absorbance needed for good UV-protection, and most efficient sunscreen compositions are desirable. The presence of scattering particles can increase the efficacy of dissolved UV-absorbers in the oil or water phases of the formulation. As many sunscreens contain UV-absorbing particles, it is of interest how much the scattering effect of such materials contribute to the protection of the respective sunscreen. The currently available software programs for simulating sunscreen performance are based on a Beer–Lambert law approach and do not take into account such scattering effects of particles. However, Monte Carlo simulations of the UV-light transport through sunscreen films are capable to take scattering from particles into consideration. Using Monte Carlo simulations, this work shows that the efficacy of absorbance is indeed increased in the presence of scattering particles. However, this is of limited significance when the particles are UV-absorbers themselves.Graphical abstract

Physics of the Signal Formation in Photopletysmography: Assessment of the Contribution of Light Absorption and Scattering to the Registered Flux of Optical Radiation

The paper is devoted to the study of physical mechanisms of photoplethysmography (PPG) signal formation using Monte Carlo simulations of light transport in biological tissue. The problem of estimating the contribution of absorption and scattering variations to the registered PPG signal is solved. Based on a three-layer skin model, changes in the optical properties of the dermal layer (absorption and scattering) were sequentially simulated and their contributions to the total signal were estimated. Calculations were carried out for two wavelengths, 525 nm and 810 nm. It was found that for green light the main contribution to the signal formation is made by absorption (88 % versus 12 % scattering, respectively). While for the near infrared light, scattering predominates over absorption. In this case, the contributions of absorption and scattering are 28 % and 72 %. Thus, for the green wavelength range the classical volumetric model of signal formation is valid. Whereas for the near-infrared range, the predominant factor in signal formation is scattering of the medium, which can change due to processes such as changes in orientation, aggregation and deformation of red blood cells, their concentration in the diagnostic volume of tissue, etc.

Technique for enhancing the accuracy of the Rayleigh-Sommerfeld convolutional diffraction through the utilization of independent spatial sampling.

The Rayleigh-Sommerfeld diffraction integral (RSD) is a rigorous solution that precisely satisfies both Maxwell's equations and Helmholtz's equations. It seamlessly integrates Huygens' principle, providing an accurate description of the coherent light propagation within the entire diffraction field. Therefore, the rapid and precise computation of the RSD is crucial for light transport simulation and optical technology applications based on it. However, the current FFT-based Rayleigh-Sommerfeld integral convolution algorithm (CRSD) exhibits poor performance in the near field, thereby limiting its applicability and impeding further development across various fields. The present study proposes, to our knowledge, a novel approach to enhance the accuracy of the Rayleigh-Sommerfeld convolution algorithm by employing independent sampling techniques in both spatial and frequency domains. The crux of this methodology involves segregating the spatial and frequency domains, followed by autonomous sampling within each domain. The proposed method significantly enhances the accuracy of RSD during the short distance while ensuring computational efficiency.

Selective photothermolysis in acne treatment: The impact of laser power.

Selective photothermolysis (SPT) using a 1726 nm laser has emerged as a safe and effective treatment option for acne vulgaris by targeting sebaceous glands (SG). Power output plays a crucial role in determining treatment selectivity and efficacy. This work highlights the advantages of a higher-power laser source and outlines the limitations of lower-power laser sources and the subsequent impact on treatment. Light transport and bioheat transfer simulations were performed to demonstrate photothermal impact on the SG and the surrounding dermis when irradiated by a high- or lower-power laser source. The simulations showed that a single higher-power-shorter-pulse (HPSP) selectively increases SG temperature well beyond bulk temperatures, which is desirable for SPT. Selectivity decreases linearly with power for the single lower-power-longer-pulses (LPLP) exposure. A multiple-LPLP approach elevates bulk temperatures significantly more than a single-pulse strategy, compromising selectivity. The goal of SPT is to damage SG safely and effectively by creating an intense temperature rise localized to the SG while moderately increasing the dermis temperature. This goal is mostly achieved with higher-power lasers that deliver a single HPSP. Lower-power lasers, longer pulse widths, and multi-pulse strategies result in higher bulk temperatures and lower SG selectivity, making such treatment challenging to execute while adding a higher risk of discomfort and downtime.

Neu(t)ralMC: energy-efficient open source Monte Carlo algorithm for assessing photon transport in turbid media.

Light propagation in turbid mediums such as atmosphere, fluids, and biological tissues is a challenging problem which necessitates accurate simulation techniques to account for the effects of multiple scattering. The Monte Carlo method has long established itself as a gold standard and is widely adopted for simulating light transport, however, its computationally intensive nature often requires significant processing power and energy consumption. In this paper a novel, open source Monte Carlo algorithm is introduced which is specifically designed for use with energy-efficient processors, effectively addressing those challenges, while maintaining the accuracy/compatibility and outperforming existing solutions. The proposed implementation optimizes photon transport simulations by exploiting the unique capabilities of Apple's low-power, high-performance M-family of chips. The developed method has been implemented in an open-source software package, enabling seamless adaptation of developed algorithms for specific applications. The accuracy and performance are validated using comprehensive comparison with existing solvers commonly used for biomedical imaging. The results demonstrate that the new algorithm achieves comparable accuracy levels to those of existing techniques while significantly reducing computational time and energy consumption.

A Practical Walk-on-Boundary Method for Boundary Value Problems

We introduce the walk-on-boundary (WoB) method for solving boundary value problems to computer graphics. WoB is a grid-free Monte Carlo solver for certain classes of second order partial differential equations. A similar Monte Carlo solver, the walk-on-spheres (WoS) method, has been recently popularized in computer graphics due to its advantages over traditional spatial discretization-based alternatives. We show that WoB's intrinsic properties yield further advantages beyond those of WoS. Unlike WoS, WoB naturally supports various boundary conditions (Dirichlet, Neumann, Robin, and mixed) for both interior and exterior domains. WoB builds upon boundary integral formulations, and it is mathematically more similar to light transport simulation in rendering than the random walk formulation of WoS. This similarity between WoB and rendering allows us to implement WoB on top of Monte Carlo ray tracing, and to incorporate advanced rendering techniques (e.g., bidirectional estimators with multiple importance sampling, the virtual point lights method, and Markov chain Monte Carlo) into WoB. WoB does not suffer from the intrinsic bias of WoS near the boundary and can estimate solutions precisely on the boundary. Our numerical results highlight the advantages of WoB over WoS as an attractive alternative to solve boundary value problems based on Monte Carlo.

A Hyperspectral Space of Skin Tones for Inverse Rendering of Biophysical Skin Properties

AbstractWe present a method for estimating the main properties of human skin, leveraging a hyperspectral dataset of skin tones synthetically generated through a biophysical layered skin model and Monte Carlo light transport simulations. Our approach learns the mapping between the skin parameters and diffuse skin reflectance in such space through an encoder‐decoder network. We assess the performance of RGB and spectral reflectance up to 1 μm, allowing the model to retrieve visible and near‐infrared. Instead of restricting the parameters to values in the ranges reported in medical literature, we allow the model to exceed such ranges to gain expressiveness to recover outliers like beard, eyebrows, rushes and other imperfections. The continuity of our albedo space allows to recover smooth textures of skin properties, enabling reflectance manipulations by meaningful edits of the skin properties. The space is robust under different illumination conditions, and presents high spectral similarity with the current largest datasets of spectral measurements of real human skin while expanding its gamut.

Anderson localization of electromagnetic waves in three dimensions

Anderson localization is a halt of diffusive wave propagation in disordered systems. Despite extensive studies over the past 40 years, Anderson localization of light in three dimensions has remained elusive, leading to the question of its very existence. Recent advances have enabled finite-difference time-domain calculations to be sped up by orders of magnitude, allowing us to conduct brute-force numerical simulations of light transport in fully disordered three-dimensional systems with unprecedented dimension and refractive index difference. We show numerically three-dimensional localization of vector electromagnetic waves in random aggregates of overlapping metallic spheres, in sharp contrast to the absence of localization for dielectric spheres with a refractive index up to 10 in air. Our work opens a wide range of avenues in both fundamental research related to Anderson localization and potential applications using three-dimensional localized light. Whether Anderson localization of light can be achieved in three dimensions has remained an open question. Numerical calculations now show that it is possible with a random arrangement of metallic spheres, but not with dielectric ones.

Validation of IES-VE for Assessing Daylight Performance of Building Implementing Horizontal Light Pipe and Shading Systems in the Tropics

Abstract. Various daylight simulation tools, which are rapidly developed, become a reliable way for simulating the complex daylighting environment. Empirical validation of the daylight simulation tool is essential in determining its reliability, especially in simulating light transport and shading system in the Tropics. Distinct from previous research, the validation involves Horizontal Light Pipe (HLP), a side window with shading systems in different room aspect ratios and orientations. This study aims to validate the simulation results of Integrated Environmental Solutions-Virtual Environment (IES-VE) Radiance IES with the measurement results of physical scaled models for evaluating HLP, light shelves, and blinds' daylight performance under intermediate and overcast sky conditions. Two physical scaled models 1:10 represent office rooms with HLP and shading systems with different room aspect ratios were constructed. Daylight Factor (DF) and Daylight Ratio (DF) of physical scaled model measurement and IES-VE simulation were compared. The results showed that under intermediate and overcast sky conditions, the Pearson correlation between simulation and measurement results using DR and DF was strong, significant, and positive, as high as 0.84 and 0.80, respectively. The Mean Bias Error between simulation and measurement results under intermediate and overcast sky conditions were -12% and -7.7%, respectively. IES-VE is reliable to evaluate the HLP and shading systems' daylight performance with different room orientations in the Tropics.

Subspace Culling for Ray-Box Intersection

Ray tracing is an essential operation for realistic image synthesis. The acceleration of ray tracing has been studied for a long period of time because algorithms such as light transport simulations require a large amount of ray tracing. One of the major approaches to accelerate the intersections is to use bounding volumes for early pruning for primitives in the volume. The axis-aligned bounding box is a popular bounding volume for ray tracing because of its simplicity and efficiency. However, the conservative bounding volume may produce extra empty space in addition to its content. Especially, primitives that are thin and diagonal to the axis give false-positive hits on the box volume due to the extra space. Although more complex bounding volumes such as oriented bounding boxes may reduce more false-positive hits, they are computationally expensive. In this paper, we propose a novel culling approach to reduce false-positive hits for the bounding box by embedding a binary voxel data structure to the volume. As a ray is represented as a conservative voxel volume as well in our approach, the ray--voxel intersection is cheaply done by bitwise AND operations. Our method is applicable to hierarchical data structures such as bounding volume hierarchy (BVH). It reduces false-positive hits due to the ray--box test and reduces the number of intersections during the traversal of BVH in ray tracing. We evaluate the reduction of intersections with several scenes and show the possibility of performance improvement despite the culling overhead. We also introduce a compression approach with a lookup table for our voxel data. We show that our compressed voxel data achieves significant false-positive reductions with a small amount of memory.

Two-term scattering phase function for photon transport to model subdiffuse reflectance in superficial tissues.

For Monte Carlo simulations of light transport in a variety of diffuse scattering applications, a single-scattering two-term phase function with five adjustable parameters is sufficiently flexible to separately control the forward and backward components of scattering. The forward component dominates light penetration into a tissue and the resulting diffuse reflectance. The backward component controls early subdiffuse scatter from superficial tissues. The phase function consists of a linear combination of two phase functions [Reynolds and McCormick, J. Opt. Soc. Am.70, 1206 (1980)10.1364/JOSA.70.001206] that were derived from the generating function for Gegenbauer polynomials. The two-term phase function (TT) accommodates strongly-forward anisotropic scattering with enhanced backscattering and is a generalization of the two-term, three-parameter Henyey-Greenstein phase function. An analytical inverse of the cumulative distribution function for scattering is provided for implementation in Monte Carlo simulations. Explicit TT equations are given for the single-scattering metrics g 1, g 2, γ, and δ. Scattering data from previously published bio-optical data are shown to fit better with the TT than other phase function models. Example Monte Carlo simulations illustrate the use of the TT and its independent control of subdiffuse scatter.

Time-Division Multiplexing Light Field Display with Learned Coded Aperture.

Conventional stereoscopic displays suffer from vergence-accommodation conflict and cause visual fatigue. Integral-imaging-based displays resolve the problem by directly projecting the sub-aperture views of a light field into the eyes using a microlens array or a similar structure. However, such displays have an inherent trade-off between angular and spatial resolutions. In this paper, we propose a novel coded time-division multiplexing technique that projects encoded sub-aperture views to the eyes of a viewer with correct cues for vergence-accommodation reflex. Given sparse light field sub-aperture views, our pipeline can provide a perception of high-resolution refocused images with minimal aliasing by jointly optimizing the sub-aperture views for display and the coded aperture pattern. This is achieved via deep learning in an end-to-end fashion by simulating light transport and image formation with Fourier optics. To our knowledge, this work is among the first that optimize the light field display pipeline with deep learning. We verify our idea with objective image quality metrics (PSNR, SSIM, and LPIPS) and perform an extensive study on various customizable design variables in our display pipeline. Experimental results show that light fields displayed using the proposed technique indeed have higher quality than that of baseline display designs.

A novel 1726-nm laser system for safe and effective treatment of acne vulgaris.

Selective photothermolysis of the sebaceous glands has the potential to be an effective alternative for treating acne vulgaris. However, the translation of this technique to clinical settings has been hindered by a lack of appropriate energy sources to target sebaceous glands, concerns surrounding safety, and treatment-related discomfort and downtime. In this work, we introduce the first FDA-approved system that combines a 1726-nm laser and efficient contact cooling to treat mild, moderate, and severe acne effectively while ensuring safety and minimal patient discomfort without adjunct pain mitigation techniques. Light transport and bioheat transfer simulations were performed to demonstrate the system's efficacy and selectivity. The resulting thermal damage to the skin and sebaceous glands was modeled using the Arrhenius kinetic model. Numerical simulations demonstrated that combining laser energy and optimal contact cooling could induce a significant temperature increase spatially limited to the sebaceous gland; this results in highly selective targeting and maximum damage to the sebaceous gland while preserving other skin structures. In vivo human facial skin histology results corroborated the simulation results. The studies reported here demonstrate that the presented 1726-nm laser system induces selective photothermolysis of the sebaceous gland, providing a safe and effective method for the treatment of acne vulgaris.

Transport through Amorphous Photonic Materials with Localization and Bandgap Regimes.

We propose a framework that unifies the description of light transmission through three-dimensional amorphous dielectric materials that exhibit both localization and a photonic bandgap. We argue that direct, coherent reflection near and in the bandgap attenuates the generation of diffuse or localized photons. Using the self-consistent theory of localization and considering the density of states of photons, we can quantitatively describe the total transmission of light for all transport regimes: transparency, light diffusion, localization, and bandgap. Comparison with numerical simulations of light transport through hyperuniform networks supports our theoretical approach.

Non-line-of-sight transient rendering

The capture and analysis of light in flight, or light in transient state, has enabled applications such as range imaging, reflectance estimation and especially non-line-of-sight (NLOS) imaging. For this last case, hidden geometry can be reconstructed using time-resolved measurements of indirect diffuse light emitted by a laser. Transient rendering is a key tool for developing such new applications, significantly more challenging than its steady-state counterpart. In this work, we introduce a set of simple yet effective subpath sampling techniques targeting transient light transport simulation in occluded scenes. We analyze the usual capture setups of NLOS scenes, where both the camera and light sources are focused on particular points in the scene. Also, the hidden geometry can be difficult to sample using conventional techniques. We leverage that configuration to reduce the integration path space. We implement our techniques in a modified version of Mitsuba 2 adapted for transient light transport, allowing us to support parallelization, polarization, and differentiable rendering.

Efficiency-aware multiple importance sampling for bidirectional rendering algorithms

Multiple importance sampling (MIS) is an indispensable tool in light-transport simulation. It enables robust Monte Carlo integration by combining samples from several techniques. However, it is well understood that such a combination is not always more efficient than using a single sampling technique. Thus a major criticism of complex combined estimators, such as bidirectional path tracing, is that they can be significantly less efficient on common scenes than simpler algorithms like forward path tracing. We propose a general method to improve MIS efficiency: By cheaply estimating the efficiencies of various technique and sample-count combinations, we can pick the best one. The key ingredient is a numerically robust and efficient scheme that uses the samples of one MIS combination to compute the efficiency of multiple other combinations. For example, we can run forward path tracing and use its samples to decide which subset of VCM to enable, and at what sampling rates. The sample count for each technique can be controlled per-pixel or globally. Applied to VCM, our approach enables robust rendering of complex scenes with caustics, without compromising efficiency on simpler scenes.