Recent Advances in Adaptive Sampling and Reconstruction for Monte Carlo Rendering

Monte Carlo integration is firmly established as the basis for most practical realistic image synthesis algorithms because of its flexibility and generality. However, the visual quality of rendered images often suffers from estimator variance, which appears as visually distracting noise. Adaptive sampling and reconstruction algorithms reduce variance by controlling the sampling density and aggregating samples in a reconstruction step, possibly over large image regions. In this paper we survey recent advances in this area. We distinguish between “a priori” methods that analyze the light transport equations and derive sampling rates and reconstruction filters from this analysis, and “a posteriori” methods that apply statistical techniques to sets of samples to drive the adaptive sampling and reconstruction process. They typically estimate the errors of several reconstruction filters, and select the best filter locally to minimize error. We discuss advantages and disadvantages of recent state-of-the-art techniques, and provide visual and quantitative comparisons. Some of these techniques are proving useful in real-world applications, and we aim to provide an overview for practitioners and researchers to assess these approaches. In addition, we discuss directions for potential further improvements.

Image Space Adaptive Rendering

In this thesis, we develop an adaptive framework for Monte Carlo rendering, and more specifically for Monte Carlo Path Tracing (MCPT) and its derivatives. MCPT is attractive because it can handle a wide variety of light transport effects, such as depth of field, motion blur, indirect illumination, participating media, and others, in an elegant and unified framework. However, MCPT is a sampling-based approach, and is only guaranteed to converge in the limit, as the sampling rate grows to infinity. At finite sampling rates, MCPT renderings are often plagued by noise artifacts that can be visually distracting. The adaptive framework developed in this thesis leverages two core strategies to address noise artifacts in renderings: adaptive sampling and adaptive reconstruction. Adaptive sampling consists in increasing the sampling rate on a per pixel basis, to ensure that each pixel value is below a predefined error threshold. Adaptive reconstruction leverages the available samples on a per pixel basis, in an attempt to have an optimal trade-off between minimizing the residual noise artifacts and preserving the edges in the image. In our framework, we greedily minimize the relative Mean Squared Error (rMSE) of the rendering by iterating over sampling and reconstruction steps. Given an initial set of samples, the reconstruction step aims at producing the rendering with the lowest rMSE on a per pixel basis, and the next sampling step then further reduces the rMSE by distributing additional samples according to the magnitude of the residual rMSE of the reconstruction. This iterative approach tightly couples the adaptive sampling and adaptive reconstruction strategies, by ensuring that we only sample densely regions of the image where adaptive reconstruction cannot properly resolve the noise. In a first implementation of our framework, we demonstrate the usefulness of our greedy error minimization using a simple reconstruction scheme leveraging a filterbank of isotropic Gaussian filters. In a second implementation, we integrate a powerful edge aware filter that can adapt to the anisotropy of the image. Finally, in a third implementation, we leverage auxiliary feature buffers that encode scene information (such as surface normals, position, or texture), to improve the robustness of the reconstruction in the presence of strong noise.

The influence of sampling design on tree-ring-based quantification of forest growth

Tree-rings offer one of the few possibilities to empirically quantify and reconstruct forest growth dynamics over years to millennia. Contemporaneously with the growing scientific community employing tree-ring parameters, recent research has suggested that commonly applied sampling designs (i.e. how and which trees are selected for dendrochronological sampling) may introduce considerable biases in quantifications of forest responses to environmental change. To date, a systematic assessment of the consequences of sampling design on dendroecological and-climatological conclusions has not yet been performed. Here, we investigate potential biases by sampling a large population of trees and replicating diverse sampling designs. This is achieved by retroactively subsetting the population and specifically testing for biases emerging for climate reconstruction, growth response to climate variability, long-term growth trends, and quantification of forest productivity. We find that commonly applied sampling designs can impart systematic biases of varying magnitude to any type of tree-ring-based investigations, independent of the total number of samples considered. Quantifications of forest growth and productivity are particularly susceptible to biases, whereas growth responses to short-term climate variability are less affected by the choice of sampling design. The world's most frequently applied sampling design, focusing on dominant trees only, can bias absolute growth rates by up to 459% and trends in excess of 200%. Our findings challenge paradigms, where a subset of samples is typically considered to be representative for the entire population. The only two sampling strategies meeting the requirements for all types of investigations are the (i) sampling of all individuals within a fixed area; and (ii) fully randomized selection of trees. This result advertises the consistent implementation of a widely applicable sampling design to simultaneously reduce uncertainties in tree-ring-based quantifications of forest growth and increase the comparability of datasets beyond individual studies, investigators, laboratories, and geographical boundaries.

Timing of bacterial carriage sampling in vaccine trials: A modelling study.

BACKGROUND Pathogenic bacteria are often asymptomatically carried in the nasopharynx. Bacterial carriage can be reduced by vaccination and has been used as an alternative endpoint to clinical disease in randomised controlled trials (RCTs). Vaccine efficacy (VE) is usually calculated as 1 minus a measure of effect. Estimates of vaccine efficacy from cross-sectional carriage data collected in RCTs are usually based on prevalence odds ratios (PORs) and prevalence ratios (PRs), but it is unclear when these should be measured. METHODS We developed dynamic compartmental transmission models simulating RCTs of a vaccine against a carried pathogen to investigate how VE can best be estimated from cross-sectional carriage data, at which time carriage should optimally be assessed, and to which factors this timing is most sensitive. In the models, vaccine could change carriage acquisition and clearance rates (leaky vaccine); values for these effects were explicitly defined (facq, 1/fdur). POR and PR were calculated from model outputs. Models differed in infection source: other participants or external sources unaffected by the trial. Simulations using multiple vaccine doses were compared to empirical data. RESULTS The combined VE against acquisition and duration calculated using POR (VEˆacq.dur, (1-POR)×100) best estimates the true VE (VEacq.dur, (1-facq×fdur)×100) for leaky vaccines in most scenarios. The mean duration of carriage was the most important factor determining the time until VEˆacq.dur first approximates VEacq.dur: if the mean duration of carriage is 1-1.5 months, up to 4 months are needed; if the mean duration is 2-3 months, up to 8 months are needed. Minor differences were seen between models with different infection sources. In RCTs with shorter intervals between vaccine doses it takes longer after the last dose until VEˆacq.dur approximates VEacq.dur. CONCLUSION The timing of sample collection should be considered when interpreting vaccine efficacy against bacterial carriage measured in RCTs.