Automatic Feature Selection for Denoising Volumetric Renderings

Zhang, Xianyao; Ott, Melvin; Manzi, Marco; Groß, Markus; Papas, Marios

doi:10.1111/cgf.14587

Cited by 3 publications

(2 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent advances in physically based modeling mainly focus on reducing image noises via producing high-quality Monte Carlo samples [13,79] or neural denoising guided by physically meaningful features [147], simulating complex material appearances such as scratches [128], caustics, and glints [146,135], representing the wave nature of light globally instead of locally [114], or simulating richer domains of natural phenomena. All in all, physically based modeling has achieved great photorealism that was never seen before.…”

Section: Introduction 11 Motivationmentioning

confidence: 99%

Chemomechanical simulation of soap film flow on spherical bubbles

et al. 2020

View full text Add to dashboard Cite

Soap bubbles are widely appreciated for their fragile nature and their colorful appearance. The natural sciences and, in extension, computer graphics, have comprehensively studied the mechanical behavior of films and foams, as well as the optical properties of thin liquid layers. In this paper, we focus on the dynamics of material flow within the soap film, which results in fascinating, extremely detailed patterns. This flow is characterized by a complex coupling between surfactant concentration and Marangoni surface tension. We propose a novel chemomechanical simulation framework rooted in lubrication theory, which makes use of a custom semi-Lagrangian advection solver to enable the simulation of soap film dynamics on spherical bubbles both in free flow as well as under body forces such as gravity or external air flow. By comparing our simulated outcomes to videos of real-world soap bubbles recorded in a studio environment, we show that our framework, for the first time, closely recreates a wide range of dynamic effects that are also observed in experiment.

show abstract

Section: Introduction 11 Motivationmentioning

confidence: 99%

Chemomechanical simulation of soap film flow on spherical bubbles

et al. 2020

View full text Add to dashboard Cite

show abstract

“…In surface rendering, the weight contribution is primarily determined by evaluating the directional differences between surfaces using normal information. However, in volumetric rendering, there is no surface normal buffer information to guide the spatial filtering [41]. Therefore, the unique gradient information in volumetric data is utilized to guide the directional contribution of the spatial filtering max(0, ( ) ( )) g g g p g q   =…”

mentioning

confidence: 99%

Real-Time Spatiotemporal Denoising Volumetric Rendering in Three-Dimensional Visualization of Puncturing Navigation

Li,

Zhou,

Shen

et al. 2024

Preprint

View full text Add to dashboard Cite

In medical diagnosis and surgical treatment, particularly in tumor puncturing surgeries, the importance of three-dimensional visualization of medical data is increasingly recognized. Traditional two-dimensional imaging techniques are limited in spatial and depth perception. This study introduces a novel real-time spatiotemporal denoising volumetric rendering technique aimed at enhancing three-dimensional visualization in puncturing navigation systems. By analyzing existing volumetric rendering methods, a spatiotemporal filtering approach is proposed. This approach filters images rendered with one sample per pixel by calculating inter-frame motion vectors in the time domain and utilizing auxiliary features in the spatial domain. It effectively reduces the noise from Monte Carlo estimations and enhances the clarity of three-dimensional organ structures. This technique achieves real-time performance exceeding 30 Hz on commercial-grade Graphics Processing Units (GPUs). The real-time spatiotemporal denoising volumetric rendering significantly enhances the three-dimensional visualization quality in puncturing navigation systems, achieving a balance between high-quality rendering and real-time performance, meeting clinical needs. This technology also has broad application potential in medical training, surgical simulation, and remote collaboration.

show abstract

Deep Adaptive Sampling and Reconstruction Using Analytic Distributions

et al. 2022

View full text Add to dashboard Cite

We propose an adaptive sampling and reconstruction method for offline Monte Carlo rendering. Our method produces sampling maps constrained by a user-defined budget that minimize the expected future denoising error. Compared to other state-of-the-art methods, which produce the necessary training data on the fly by composing pre-rendered images, our method samples from analytic noise distributions instead. These distributions are compact and closely approximate the pixel value distributions stemming from Monte Carlo rendering. Our method can efficiently sample training data by leveraging only a few per-pixel statistics of the target distribution, which provides several benefits over the current state of the art. Most notably, our analytic distributions' modeling accuracy and sampling efficiency increase with sample count, essential for high-quality offline rendering. Although our distributions are approximate, our method supports joint end-to-end training of the sampling and denoising networks. Finally, we propose the addition of a global summary module to our architecture that accumulates valuable information from image regions outside of the network's receptive field. This information discourages sub-optimal decisions based on local information. Our evaluation against other state-of-the-art neural sampling methods demonstrates denoising quality and data efficiency improvements.

show abstract

Automatic Feature Selection for Denoising Volumetric Renderings

Cited by 3 publications

References 45 publications

Chemomechanical simulation of soap film flow on spherical bubbles

Chemomechanical simulation of soap film flow on spherical bubbles

Real-Time Spatiotemporal Denoising Volumetric Rendering in Three-Dimensional Visualization of Puncturing Navigation

Deep Adaptive Sampling and Reconstruction Using Analytic Distributions

Contact Info

Product

Resources

About