• Journal of Information Processing Systems
• /
• v.17 no.4
• /
• pp.737-753
• /
• 2021

As the demand for high-quality rendering for mixed reality, videogame, and simulation has increased, global illumination has been actively researched. Monte Carlo path tracing can realize global illumination and produce photorealistic scenes that include critical effects such as color bleeding, caustics, multiple light, and shadows. If the sampling rate is insufficient, however, the rendered results have a large amount of noise. The most successful approach to eliminating or reducing Monte Carlo noise uses a feature-based filter. It exploits the scene characteristics such as a position within a world coordinate and a shading normal. In general, the techniques are based on the denoised pixel or sample and are computationally expensive. However, the main challenge for all of them is to find the appropriate weights for every feature while preserving the details of the scene. In this paper, we compare the recent algorithms for removing Monte Carlo noise in terms of their performance and quality. We also describe their advantages and disadvantages. As far as we know, this study is the first in the world to compare the artificial intelligence-based denoising methods for Monte Carlo rendering.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1366-1372
• /
• 2019

In this paper, we designed a revised neural network to remove the Monte Carlo Rendering noise contained in the ray tracing graphics. The Monte Carlo Rendering is the best way to enhance the graphic's realism, but because of the need to calculate more than thousands of light effects per pixel, rendering processing time has increased rapidly, causing a major problem with real-time processing. To improve this problem, the number of light used in pixels is reduced, where rendering noise occurs and various studies have been conducted to eliminate this noise. In this paper, a deep learning is used to remove rendering noise, especially by separating the rendering image into diffuse and specular light, so that the structure of the dual neural network is designed. As a result, the dual neural network improved by an average of 0.58 db for 64 test images based on PSNR, and 99.22% less light compared to reference image, enabling real-time race-tracing rendering.

• Nuclear Engineering and Technology
• /
• v.41 no.3
• /
• pp.279-286
• /
• 2009

Improving over a previous study [1], this paper provides a Monte Carlo method for the heat conduction analysis of problems with complicated geometry (such as a pebble with dispersed fuel particles). The method is based on the theoretical results of asymptotic analysis of neutron transport equation. The improved method uses an appropriate boundary layer correction (with extrapolation thickness) and a scaling factor, rendering the problem more diffusive and thus obtaining a heat conduction solution. Monte Carlo results are obtained for the randomly distributed fuel particles of a pebble, providing realistic temperature distributions (showing the kernel and graphite-matrix temperatures distinctly). The volumetric analytic solution commonly used in the literature is shown to predict lower temperatures than those of the Monte Carlo results provided in this paper.

• Nuclear Engineering and Technology
• /
• v.49 no.6
• /
• pp.1189-1198
• /
• 2017

Computer-aided design (CAD)-based Monte Carlo radiation transport is of value to the nuclear engineering community for its ability to conduct transport on high-fidelity models of nuclear systems, but it is more computationally expensive than native geometry representations. This work describes the adaptation of a rendering data structure, the signed distance field, as a geometric query tool for accelerating CAD-based transport in the direct-accelerated geometry Monte Carlo toolkit. Demonstrations of its effectiveness are shown for several problems. The beginnings of a predictive model for the data structure's utilization based on various problem parameters is also introduced.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.10 no.1
• /
• pp.381-394
• /
• 2016

The several noise level estimation algorithms that have been developed for use in image processing and computer graphics generally exhibit good performance. However, there are certain special types of noisy images that such algorithms are not suitable for. It is particularly still a challenge to use the algorithms to estimate the noise levels of complex textured photographic images because of the inhomogeneity of the original scenes. Similarly, it is difficult to apply most conventional noise level estimation algorithms to images rendered by the Monte Carlo (MC) method owing to the spatial variation of the noise in such images. This paper proposes a novel noise level estimation method based on histogram modification, and which can be used for more accurate estimation of the noise levels in both complex textured images and MC-rendered images. The proposed method has good performance, is simple to implement, and can be efficiently used in various image-based and graphic applications ranging from smartphone camera noise removal to game background rendition.

• Journal of KIISE
• /
• v.43 no.8
• /
• pp.835-840
• /
• 2016

Monte Carlo ray tracing has been widely used for simulating a diverse set of photo-realistic effects. However, this technique typically produces noise when insufficient numbers of samples are used. As the number of samples allocated per pixel is increased, the rendered images converge. However, this approach of generating sufficient numbers of samples, requires prohibitive rendering time. To solve this problem, image filtering can be applied to rendered images, by filtering the noisy image rendered using low sample counts and acquiring smoothed images, instead of naively generating additional rays. In this paper, we proposed a Stein's Unbiased Risk Estimator (SURE) based $\grave{A}$-Trous wavelet to filter the noise in rendered images in a near-interactive rate. Based on SURE, we can estimate filtering errors associated with $\grave{A}$-Trous wavelet, and identify wavelet coefficients reducing filtering errors. Our approach showed improvement, up to 6:1, over the original $\grave{A}$-Trous filter on various regions in the image, while maintaining a minor computational overhead. We have integrated our propsed filtering method with the recent interactive ray tracing system, Embree, and demonstrated its benefits.

• Journal of the Korea Computer Graphics Society
• /
• v.16 no.3
• /
• pp.1-10
• /
• 2010

In this paper, we propose a novel noiseless method of BRDF rendering on a GPU in real-time. Illumination at a surface point is formulated as an integral of BRDF producted with incident radiance over the hemi-sphere domain. The most popular method to compute the integral is the Monte Carlo method, which needs a large number of samples to achieve good image quality. But, it leads to increase of rendering time. Otherwise, a small number of sample points cause serious image noise. The main contribution of our work is a new importance sampling scheme producing a set of incoming ray samples varying continuously with respect to the eye ray. An incoming ray is importance-based sampled at different latitude angles of the eye ray, and then the ray samples are linearly connected to form a curve, called a thread. These threads give continuously moving incident rays for eye ray change, so they do not make image noise. Since even a small number of threads can achieve a plausible quality and also can be precomputed before rendering, they enable real-time BRDF rendering on the GPU.

• Annual Conference of KIPS
• /
• 2021.11a
• /
• pp.960-962
• /
• 2021

Monte Carlo 렌더링은 모든 빛을 광원에서부터 추적하는 것 대신, 몇 개의 빛의 경로만을 추적해서 이들의 평균으로 화소값을 정해 이미지를 만드는 방법이다. 여기서 추적하는 빛이 많다면 이미지가 사실적으로 만들어질 수 있지만 연산량이 증가한다. 따라서 적은 빛의 경로를 추적하여 렌더링을 수행하여 이미지를 만들고, 노이즈를 제거해서 많은 양의 빛을 추적하여 렌더링을 한 이미지와 유사하게 만들려는 연구가 많이 진행되고 있다. 그러나 이러한 연구들은 많은 연산량을 요구하기 때문에 고성능의 기기 사양을 요구한다. 따라서 본 연구에서는 저사양의 기기에서 활용할 수 있도록 Harris corner 검출법과 median filtering을 활용한 렌더링 이미지 노이즈 제거 연구를 수행했다.

• Journal of Korea Game Society
• /
• v.7 no.2
• /
• pp.21-32
• /
• 2007

It is significant to simulate scattering of light within media for realistic image synthesis; however, this requires costly computation. This paper introduces a practical image-space approximation technique for interactive subsurface scattering. We use a general two-pass approach, which creates transmitted irradiance samples onto shadow maps and computes illumination using the shadow maps. We estimate single scattering efficiently using a method similar to common shadow mapping with adaptive deterministic sampling. A hierarchical technique is applied to evaluate multiple scattering, based on a diffusion theory. We further accelerate rendering speed by tabulating complex functions and utilizing level of detail. We demonstrate that our technique produces high-quality images of animated scenes with blurred shadow at hundreds frames per second on graphics hardware. It can be integrated into existing interactive systems easily.

• Journal of The Korean Astronomical Society
• /
• v.51 no.1
• /
• pp.5-16
• /
• 2018

We investigate the escape of $Ly{\beta}$ from emission nebulae with a significant population of excited hydrogen atoms in the level n = 2, rendering them optically thick in $H{\alpha}$. The transfer of $Ly{\beta}$ line photons in these optically thick regions is complicated by the presence of another scattering channel leading to re-emission of $H{\alpha}$, alternating their identities between $Ly{\beta}$ and $H{\alpha}$. In this work, we develop a Monte Carlo code to simulate the transfer of $Ly{\beta}$ line photons incorporating the scattering channel into $H{\alpha}$. Both $H{\alpha}$ and $Ly{\beta}$ lines are formed through diffusion in frequency space, where a line photon enters the wing regime after a fairly large number of resonance scatterings with hydrogen atoms. Various line profiles of $H{\alpha}$ and $Ly{\beta}$ emergent from our model nebulae are presented. It is argued that the electron temperature is a critical parameter which controls the flux ratio of emergent $Ly{\beta}$ and $H{\alpha}$. Specifically for $T\;=\;3{\times}10^4\;K$ and $H{\alpha}$ line center optical depth $\tau{\alpha}\;=\;10$, the number flux ratio of emergent $Ly{\beta}$ and $H{\alpha}$ is ~ 49 percent, which is quite significant. We propose that the leaking $Ly{\beta}$ can be an interesting source for the formation of $H{\alpha}$ wings observed in many symbiotic stars and active galactic nuclei. Similar broad $H{\alpha}$ wings are also expected in $Ly{\alpha}$ emitting halos found in the early universe, which can be potentially probed by the James Webb Telescope in the future.