28 Hits in 7.8 sec

Sample-based Monte Carlo denoising using a kernel-splatting network

Michaël Gharbi, Tzu-Mao Li, Miika Aittala, Jaakko Lehtinen, Frédo Durand
2019 ACM Transactions on Graphics  
State-of-the-art pixel-based Monte Carlo denoising algorithms (right) struggle with very noisy inputs rendered with a low sample count (left).  ...  We present the first convolutional network that can learn to denoise Monte Carlo renderings directly from the samples.  ...  SAMPLE-BASED DENOISING NETWORK We cast Monte Carlo denoising as a supervised learning problem.  ... 
doi:10.1145/3306346.3322954 fatcat:guikcr2fefc4xolmktbo5yei4m

Real‐time Monte Carlo Denoising with Weight Sharing Kernel Prediction Network

Hangming Fan, Rui Wang, Yuchi Huo, Hujun Bao
2021 Computer graphics forum (Print)  
Recently, kernel-prediction methods use a neural network to predict each pixel's filtering kernel and have shown a great potential to remove Monte Carlo noise.  ...  Real-time Monte Carlo denoising aims at removing severe noise under low samples per pixel (spp) in a strict time budget.  ...  [MH20]used the hierarchical kernel prediction architecture to denoise the re-sampled Monte Carlo images and the samples-splatted layers, respectively, and they achieved an interactive speed.Besides, Thomas  ... 
doi:10.1111/cgf.14338 fatcat:2nky2b3ywrfphoemacddqdb77m

Real-time Monte Carlo Denoising with the Neural Bilateral Grid

Xiaoxu Meng, Quan Zheng, Amitabh Varshney, Gurprit Singh, Matthias Zwicker
2020 Eurographics Symposium on Rendering  
In this paper, we propose a novel and practical deep learning approach to robustly denoise Monte Carlo images rendered at sampling rates as low as a single sample per pixel (1-spp).  ...  At such low sampling rates, our approach outperforms state-of-the-art techniques based on kernel prediction networks both in terms of quality and speed, and it leads to significantly improved quality compared  ...  In this paper, we propose a novel deep learning approach to denoise Monte Carlo images rendered with extremely few samples.  ... 
doi:10.2312/sr.20201133 dblp:conf/rt/MengZVSZ20 fatcat:qpzyz3eihfgmfmp6ko7zgbtmhy

Evaluation of Artificial Intelligence-Based Denoising Methods for Global Illumination

Soroor Malekmohammadi Faradounbeh, SeongKi Kim
2021 Journal of Information Processing Systems  
The most successful approach to eliminating or reducing Monte Carlo noise uses a feature-based filter.  ...  In general, the techniques are based on the denoised pixel or sample and are computationally expensive.  ...  Sample-based Monte Carlo denoising (SBMCD) A convolutional network can learn to denoise MC renderings directly from the samples.  ... 
doi:10.3745/jips.02.0162 dblp:journals/jips/FaradounbehK21 fatcat:s4fw6suisncmneiyeovh2kqie4

A survey on deep learning-based Monte Carlo denoising

Yuchi Huo, Sung-eui Yoon
2021 Computational Visual Media  
Recent years have seen increasing attention and significant progress in denoising MC rendering with deep learning, by training neural networks to reconstruct denoised rendering results from sparse MC samples  ...  AbstractMonte Carlo (MC) integration is used ubiquitously in realistic image synthesis because of its flexibility and generality.  ...  Fig. 1 1 Deep learning-based Monte Carlo denoising method trains a neural network to reduce Monte Carlo noise in input images. Reproduced with permission from Ref. [9] , c Author 2018.  ... 
doi:10.1007/s41095-021-0209-9 fatcat:pki3mbbpw5fjxf6oaphcnlto4a

DGAN-KPN: Deep Generative Adversarial Network and Kernel Prediction Network for Denoising MC Renderings

Ahmed Mustafa Taha Alzbier, Chunyi Chen
2022 Symmetry  
In this paper, we present a denoising network composed of a kernel prediction network and a deep generative adversarial network to construct an end-to-end overall network structure.  ...  The generator model takes the noisy Monte Carlo-rendered image as the input, passes through the symmetric encoder–decoder structure and the residual block structure, and finally outputs the rendered image  ...  In addition, the denoising effects of the data rendered by multiple renderers showed that the network model of this paper had a relatively good generalization ability and a good adaptability to the rendering  ... 
doi:10.3390/sym14020395 fatcat:igyh45t4rva7dbxu3rvi6dadwa

Fast Monte Carlo Rendering via Multi-Resolution Sampling [article]

Qiqi Hou, Zhan Li, Carl S Marshall, Selvakumar Panneer, Feng Liu
2021 arXiv   pre-print
Monte Carlo rendering algorithms are widely used to produce photorealistic computer graphics images.  ...  In this paper, we present a hybrid rendering method to speed up Monte Carlo rendering algorithms.  ...  ACKNOWLEDGMENTS The source models in Figure 1 , 2, 4, 7, and 11 are used under a Creative Commons License from kujaba, darkst0ne, cczero, LukeLiptak, Christophe Seux, nickbrunner, samytichadou, MarcoD  ... 
arXiv:2106.12802v1 fatcat:qmk243d3v5c4dmmtlgfsyybrhq

Deep Photon Mapping [article]

Shilin Zhu, Zexiang Xu, Henrik Wann Jensen, Hao Su, Ravi Ramamoorthi
2020 arXiv   pre-print
Recently, deep learning-based denoising approaches have led to dramatic improvements in low sample-count Monte Carlo rendering.  ...  We train a novel deep neural network to predict a kernel function to aggregate photon contributions at shading points.  ...  [ ] make use of individual screen-space path samples and predict a kernel for each sample that splats the radiance contributions to its neighboring pixels.  ... 
arXiv:2004.12069v1 fatcat:fias4l6d5ngdnlxhtgsmwhboke

Survey: Machine Learning in Production Rendering [article]

Shilin Zhu
2020 arXiv   pre-print
This survey summarizes several of the most dramatic improvements in using deep neural networks over traditional rendering methods, such as better image quality and lower computational overhead.  ...  In the past few years, machine learning-based approaches have had some great success for rendering animated feature films.  ...  Gharbi et al. [2019] focused on individual samples and used a combined MLP and CNN network to predict a kernel that splatted the radiance contribution of each sample onto the neighboring image pixels.  ... 
arXiv:2005.12518v1 fatcat:htymxgpc2jdmre6umrlh2xvsu4

Deep Radiance Caching: Convolutional Autoencoders Deeper in Ray Tracing

Giulio Jiang, Bernhard Kainz
2020 Computers & graphics  
Sample-based [46] Mara, M, McGuire, M, Bitterli, B, Jarosz, W. An efficient de- 118 47 monte carlo denoising using a kernel-splatting network.  ...  Kernel-predicting convolutional networks for denoising monte carlo M, et al. Noise2noise: Learning image restoration without clean data. 101 30 renderings.  ... 
doi:10.1016/j.cag.2020.09.007 fatcat:qwufxhulcbaqfksbwbzfec7fv4

Sparse Sampling for Real-time Ray Tracing

Timo Viitanen, Matias Koskela, Kalle Immonen, Markku Mäkitalo, Pekka Jääskeläinen, Jarmo Takala
2018 Proceedings of the 13th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications  
Uneven sampling methods tend to require at least one sample per pixel, limiting their use in real-time rendering.  ...  We review recent work on image reconstruction from arbitrarily distributed samples, and argue that these will play major role in the future of real-time ray tracing, allowing a larger fraction of samples  ...  A separate question is how to distribute the adaptive samples. Offline adaptive ray tracers most often sample based on Monte Carlo variance from several previous samples (Zwicker et al., 2015) .  ... 
doi:10.5220/0006655802950302 dblp:conf/grapp/ViitanenKIMJT18 fatcat:4likzzmtijgw3imkvjs5jsogqm

Deep Radiance Caching: Convolutional Autoencoders Deeper in Ray Tracing [article]

Giulio Jiang, Bernhard Kainz
2020 arXiv   pre-print
DRC employs a denoising neural network with Radiance Caching to support a wide range of material types, without the requirement of offline pre-computation or training for each scene.This offers high performance  ...  Recent research uses Deep Neural Networks to predict indirect lighting on image level, but such methods are commonly limited to diffuse materials and require training on each scene.We present Deep Radiance  ...  Kernel-predicting convolutional networks for denoising Monte Carlo renderings [16] and the recent extension Denoising with Kernel Prediction and Asymmetric Loss Functions [17] improve denoising networks  ... 
arXiv:1910.02480v2 fatcat:tmikya7qdvarvbps736z3usnw4

Path Tracing Denoising based on SURE Adaptive Sampling and Neural Network

Qiwei Xing, Chunyi Chen
2020 IEEE Access  
INDEX TERMS Adaptive sampling, SURE estimator, MLPs network, path tracing, denoising.  ...  . • Compared with those methods which used the neural network to predict the kernel size of the filter, we introduce a set of additional features extracted from the path tracing rendering process, such  ...  [4] present a recent survey of denoising techniques for Monte Carlo rendering.  ... 
doi:10.1109/access.2020.2999891 fatcat:l3efvetnefbptez3ejknvucvgy

Laplacian kernel splatting for efficient depth-of-field and motion blur synthesis or reconstruction

Thomas Leimkühler, Hans-Peter Seidel, Tobias Ritschel
2018 ACM Transactions on Graphics  
At runtime, spreadlets can be splat eiciently to the Laplacian of an image. Integrating this image produces the inal result.  ...  Splatting the point-spread function (PSF) of every pixel is general and provides high quality, but requires prohibitive compute time.  ...  A typical DoF solution is to use one [Kraus and Strengert 2007] , or multiple MIP fetches [Lee et al. 2008] or learned kernels in a neural network [Nalbach et al. 2017] .  ... 
doi:10.1145/3197517.3201379 fatcat:io7vkvxbmzfbna5amar3yszpcq

Deep-learning the Latent Space of Light Transport [article]

Pedro Hermosilla and Sebastian Maisch and Tobias Ritschel and Timo Ropinski
2019 arXiv   pre-print
Thus, we suggest a two-stage operator comprising of a 3D network that first transforms the point cloud into a latent representation, which is later on projected to the 2D output image using a dedicated  ...  3D-2D network in a second step.  ...  We would like to acknowledge the NVIDIA Corporation for donating a Quadro P6000 for our training cluster, and Gloria Fackelmann for providing the voice over the supplementary video.  ... 
arXiv:1811.04756v2 fatcat:cjpkvlradra7lf5bma7wfjmnce
« Previous Showing results 1 — 15 out of 28 results