Explicit Modeling of Personal Space for Improved Local Dynamics in Simulated Crowds

Crowd simulation demands careful consideration in regard to the classic trade-off between accuracy and efficiency. Particle-based methods have seen success in various applications in architecture, military, urban planning, and entertainment. This method focuses on local dynamics of individuals in large crowds, with a focus on serious games and entertainment. The technique uses an area-based penalty force that captures the infringement of each entity's personal space. This method does not need a costly nearest-neighbor search and allows for an inherently data-parallel implementation capable of simulating thousands of entities at interactive frame rates. The algorithm reproduces personal space compression around motion barriers for moving crowds and around points of interest for static crowds.

Automatic object annotation in streamed and remotely explored large 3D reconstructions

We introduce a novel framework for 3D scene reconstruction with simultaneous object annotation, using a pre-trained 2D convolutional neural network (CNN), incremental data streaming, and remote exploration, with a virtual reality setup. It enables versatile integration of any 2D box detection or segmentation network. We integrate new approaches to (i) asynchronously perform dense 3D-reconstruction and object annotation at interactive frame rates, (ii) efficiently optimize CNN results in terms of object prediction and spatial accuracy, and (iii) generate computationally-efficient colliders in large triangulated 3D-reconstructions at run-time for 3D scene interaction. Our method is novel in combining CNNs with long and varying inference time with live 3D-reconstruction from RGB-D camera input. We further propose a lightweight data structure to store the 3D-reconstruction data and object annotations to enable fast incremental data transmission for real-time exploration with a remote client, which has not been presented before. Our framework achieves update rates of 22 fps (SSD Mobile Net) and 19 fps (Mask RCNN) for indoor environments up to 800 m3. We evaluated the accuracy of 3D-object detection. Our work provides a versatile foundation for semantic scene understanding of large streamed 3D-reconstructions, while being independent from the CNN’s processing time. Source code is available for non-commercial use.

Dynamique des rapports de places et des images identitaires dans Les parents terribles de Jean Cocteau

"Dynamics of Place Relationships and Identity Images in Les parents terribles by Jean Cocteau. The article aims to study the dynamics of the place relationships between mother and son in scene IV of the first act of the above-mentioned play. This relationship, which represents the central theme of the play, is based on authority and domination and the son will try to ward off by resorting to a strategy. We will try to see first of all what this strategy consists of and what effect it will have on the evolution of the relationships between the two characters. Then, the study of the successive phases through which verbal interaction goes through (conversation, discussion, argument) will allow us to show how the plurality of interlocutive roles influences the dynamics of place relationships and identity images. Keywords: verbal interaction, place relationships, identity images, interactive frame, interactive space, conversation, discussion, argument."

Hash-Based Hierarchical Caching and Layered Filtering for Interactive Previews in Global Illumination Rendering

Modern Monte-Carlo-based rendering systems still suffer from the computational complexity involved in the generation of noise-free images, making it challenging to synthesize interactive previews. We present a framework suited for rendering such previews of static scenes using a caching technique that builds upon a linkless octree. Our approach allows for memory-efficient storage and constant-time lookup to cache diffuse illumination at multiple hitpoints along the traced paths. Non-diffuse surfaces are dealt with in a hybrid way in order to reconstruct view-dependent illumination while maintaining interactive frame rates. By evaluating the visual fidelity against ground truth sequences and by benchmarking, we show that our approach compares well to low-noise path-traced results, but with a greatly reduced computational complexity, allowing for interactive frame rates. This way, our caching technique provides a useful tool for global illumination previews and multi-view rendering.

Hash-Based Hierarchical Caching and Layered Filtering for Interactive Previews in Global Illumination Rendering

Modern Monte-Carlo-based rendering systems still suffer from the computational complexity involved in the generation of noise-free images, making it challenging to synthesize interactive previews. We present a framework suited for rendering such previews of static scenes using a caching technique that builds upon a linkless octree. Our approach allows for memory-efficient storage and constant-time lookup to cache diffuse illumination at multiple hitpoints along the traced paths. Non-diffuse surfaces are dealt with in a hybrid way in order to reconstruct view-dependent illumination while maintaining interactive frame rates. By evaluating the visual fidelity against ground truth sequences and by benchmarking, we show that our approach compares well to low-noise path traced results, but with a greatly reduced computational complexity allowing for interactive frame rates. This way, our caching technique provides a useful tool for global illumination previews and multi-view rendering.

Uma Nova Solução para Visibility Culling de Cenas Genéricas, baseada em Replicação, Heurísticas e Redução de Draw Calls

Graphics applications with visual quality and increasing levels of interactivity have been of fundamental interest. Within this context, visibility culling algorithms restrict the processing to the objects actually visible by the observer, speeding up the scene visualization. However, state-of-the-art solutions still require a high computational cost, do not scale in complex scenarios and are limited in generalization. In contrast, this work presents RHView, an innovative generic solution for static and dynamics scenes, which is based on a replicated space-partitioning structure and heuristics. RHView uses novel heuristics for rendering time estimation and balance between processing cost and triangle removal accuracy, while maintaining interactive frame rates, even in scenes with billions of triangles. It is the only solution currently available to reduce draw calls, one of the factors that have the greatest impact on graphics processing. Systematic tests have shown that RHView can be up to 2.8 times faster than the state-of-the-art algorithms.

Ecologically Sound Procedural Generation of Natural Environments

Current techniques for the creation and exploration of virtual worlds are largely unable to generate sound natural environments from ecological data and to provide interactive web-based visualizations of such detailed environments. We tackle this challenge and propose a novel framework that (i) explores the advantages of landscape maps and ecological statistical data, translating them to an ecologically sound plant distribution, and (ii) creates a visually convincing 3D representation of the natural environment suitable for its interactive visualization over the web. Our vegetation model improves techniques from procedural ecosystem generation and neutral landscape modeling. It is able to generate diverse ecological sound plant distributions directly from landscape maps with statistical ecological data. Our visualization model integrates existing level of detail and illumination techniques to achieve interactive frame rates and improve realism. We validated with ecology experts the outcome of our framework using two case studies and concluded that it provides convincing interactive visualizations of large natural environments.

Comparison of Acceleration Data Structures for High Quality Fast Reflections of Static and Deformable Models in Walkthrough Animations

In order to render realistic images, the reflectance of surfaces must be simulated accurately. Generally, the ray tracing rendering technique is used to make a material reflect its surroundings, since it represents with great fidelity the behavior of light. However, ray tracing is still a very costly algorithm, so far mostly indicated in offline rendering scenarios. This situation is even more challenging for scenes containing 3D deformable meshes, since their geometry and, thus, the acceleration structures used, need to be updated in each frame of the animation. In this paper, we present an extended version of our hybrid algorithm that combines rasterization and a pure ray tracing through the NVIDIA OptiX to render high quality fast reflections, including scenes with deformable models. Additionally, we analyze and compare the performances of different NVIDIA OptiX acceleration data structures for generating reflections of static and deformable models in walkthrough animations. The results show that NVIDIA OptiX acceleration structures reach high frames per second for static objects. However, there is a performance decay in terms of frames per second when dealing with deformable models, since it becomes necessary to update the acceleration structures to cope with changing geometry, but even under these restrictions, we were able to achieve interactive frame rates.