EG 2019 - Short Papers

Permanent URI for this collection

https://diglib.eg.org/handle/10.2312/2632750

Browse

Now showing 1 - 13 of 13

3DVFX: 3D Video Editing using Non-Rigid Structure-from-Motion
(The Eurographics Association, 2019) parashar, shaifali; Bartoli, Adrien; Cignoni, Paolo and Miguel, Eder
Numerous video post-processing techniques can add or remove objects to the observed scene in the video. Most of these techniques rely on 2D image points to perform the desired changes. Structure-from-Motion (SfM) has allowed the use of 3D points, however only for the objects that remain rigid in the scene. We propose to use both 2D image points and 3D points to modify the scene's deformable objects using Non-Rigid Structure-from-Motion (NRSfM). We rely on a recent effective NRSfM solution to develop a complete pipeline including manual 3D editing of an image and automatic 3D transfer of the edits. We perform object manipulation tasks such as retexturing a real deforming object.
Anisotropic Filtering for On-the-fly Patch-based Texturing
(The Eurographics Association, 2019) Lutz, Nicolas; Sauvage, Basile; Larue, Frédéric; Dischler, Jean-Michel; Cignoni, Paolo and Miguel, Eder
On-the-fly patch-based texturing consists of choosing at run-time, for several patches within a tileable texture, one random candidate among a pre-computed set of possible contents. This category of methods generates unbounded textures, for which filtering is not straightforward, because the screen pixel footprint may overlap multiple patches in texture space, i.e. different randomly chosen contents. In this paper, we propose a real-time anisotropic filtering which is fully compliant with the standard graphics pipeline. The main idea is to pre-filter the contents independently, store them in an atlas, and combine them at run-time to produce the final pixel color. The patch-map, referencing to which patch belong the fetched texels, requires a specific filtering approach, in order to recover the patches that overlap at low resolutions. In addition, we show how this method can achieve blending at patch boundaries in order to further reduce visible seams, without modification of our filtering algorithm.
Area Lights in Signed Distance Function Scenes
(The Eurographics Association, 2019) Bán, Róbert; Bálint, Csaba; Valasek, Gábor; Cignoni, Paolo and Miguel, Eder
This paper presents two algorithms to incorporate spherical and general area lights into scenes defined by signed distance functions. The first algorithm employs an efficient approximation to the contribution of spherical lights to direct illumination and renders them at real-time rates. The second algorithm is of superior quality at a higher computational cost which is better suited for interactive rates. Our results are compared to both real-time soft shadow algorithms and a ground truth obtained by Monte Carlo integration. We show in these comparisons that our real-time solution computes more accurate shadows while the more demanding variant outperforms Monte Carlo integration at the expense of accuracy.
Fine-Grained Semantic Segmentation of Motion Capture Data using Dilated Temporal Fully-Convolutional Networks
(The Eurographics Association, 2019) Cheema, Noshaba; hosseini, somayeh; Sprenger, Janis; Herrmann, Erik; Du, Han; Fischer, Klaus; Slusallek, Philipp; Cignoni, Paolo and Miguel, Eder
Human motion capture data has been widely used in data-driven character animation. In order to generate realistic, naturallooking motions, most data-driven approaches require considerable efforts of pre-processing, including motion segmentation and annotation. Existing (semi-) automatic solutions either require hand-crafted features for motion segmentation or do not produce the semantic annotations required for motion synthesis and building large-scale motion databases. In addition, human labeled annotation data suffers from inter- and intra-labeler inconsistencies by design. We propose a semi-automatic framework for semantic segmentation of motion capture data based on supervised machine learning techniques. It first transforms a motion capture sequence into a ''motion image'' and applies a convolutional neural network for image segmentation. Dilated temporal convolutions enable the extraction of temporal information from a large receptive field. Our model outperforms two state-of-the-art models for action segmentation, as well as a popular network for sequence modeling. Most of all, our method is very robust under noisy and inaccurate training labels and thus can handle human errors during the labeling process.
Font Specificity
(The Eurographics Association, 2019) Power, Luther; Lau, Manfred; Cignoni, Paolo and Miguel, Eder
We explore the concept of ''image specificity'' for fonts and introduce the notion of ''font specificity''. The idea is that a font that elicits consistent descriptions from different people are more ''specific''. We collect specificity-based data for fonts where participants are given each font and asked to describe it with words. We then analyze the data and characterize the qualitative features that make a font ''specific''. Finally, we show that the notion of font specificity can be learned and demonstrate some specificity-guided applications.
GPU Smoke Simulation on Compressed DCT Space
(The Eurographics Association, 2019) Ishida, Daichi; Ando, Ryoichi; Morishima, Shigeo; Cignoni, Paolo and Miguel, Eder
This paper presents a novel GPU-based algorithm for smoke animation. Our primary contribution is the use of Discrete Cosine Transform (DCT) compressed space for efficient simulation. We show that our method runs an order of magnitude faster than a CPU implementation while retaining visual details with a smaller memory usage. The key component of our method is an on-the-fly compression and expansion of velocity, pressure and density fields. Whenever these physical quantities are requested during a simulation, we perform data expansion and compression only where necessary in a loop. As a consequence, our simulation allows us to simulate a large domain without actually allocating full memory space for it. We show that albeit our method comes with some extra cost for DCT manipulations, such cost can be minimized with the aid of a devised shared memory usage.
Making Gabor Noise Fast and Normalized
(The Eurographics Association, 2019) Tavernier, Vincent; Neyret, Fabrice; Vergne, Romain; Thollot, Joëlle; Cignoni, Paolo and Miguel, Eder
Gabor Noise is a powerful procedural texture synthesis technique, but it has two major drawbacks: It is costly due to the high required splat density and not always predictable because properties of instances can differ from those of the process. We bench performance and quality using alternatives for each Gabor Noise ingredient: point distribution, kernel weighting and kernel shape. For this, we introduce 3 objective criteria to measure process convergence, process stationarity, and instance stationarity. We show that minor implementation changes allow for 17-24x speed-up with same or better quality.
Perceptual Characteristics by Motion Style Category
(The Eurographics Association, 2019) Kim, Hye Ji; Lee, Sung-Hee; Cignoni, Paolo and Miguel, Eder
Motion style is important as it characterizes a motion by expressing the context of the motion such as emotion and personality. Yet, the perception and interpretation of motion styles is subjective and may vary greatly from person to person. This paper investigates the perceptual characteristics of motion styles for a wide range of styles. After categorizing the motion styles, we perform user studies to examine the diversity of interpretations of motion styles and the association level between style motions and their corresponding text descriptions. Our study shows that motion styles have different interpretation diversity and association level according to their categories. We discuss the implications of these findings and recommend a method of labeling or describing motion styles.
Planar Abstraction and Inverse Rendering of 3D Indoor Environment
(The Eurographics Association, 2019) Kim, Young Min; Ryu, Sangwoo; Kim, Ig-Jae; Cignoni, Paolo and Miguel, Eder
A large-scale scanned 3D environment suffers from complex occlusions and misalignment errors. The reconstruction contains holes in geometry and ghosting in texture. These are easily noticed and cannot be used in visually compelling VR content without further processing. On the other hand, the well-known Manhattan World priors successfully recreate relatively simple or clean structures. In this paper, we would like to push the limit of planar representation in indoor environments. We use planes not only to represent the environment geometrically but also to solve an inverse rendering problem considering texture and light. The complex process of shape inference and intrinsic imaging is greatly simplified with the help of detected planes and yet produces a realistic 3D indoor environment. The produced content can effectively represent the spatial arrangements for various AR/VR applications and can be readily combined with virtual objects possessing plausible lighting and texture.
A Preliminary Analysis of Methods for Curvature Estimation on Surfaces With Local Reliefs
(The Eurographics Association, 2019) Moscoso Thompson, Elia; Biasotti, Silvia; Cignoni, Paolo and Miguel, Eder
Curvature estimation is very popular in geometry processing for the analysis of local surface variations. Despite the large number of methods, no quantitative nor qualitative studies have been conducted for a comparative analysis of the different algorithms on surfaces with small geometric variations, such as chiselled or relief surfaces. In this work we compare eight curvature estimation methods that are commonly adopted by the computer graphics community on a number of triangle meshes derived from scans of surfaces with local reliefs.
Schelling Meshes
(The Eurographics Association, 2019) Power, Luther; Lau, Manfred; Cignoni, Paolo and Miguel, Eder
The concept of ''Schelling points'' on 3D shapes has been explored for points on the surface of a 3D mesh. In this paper, we introduce the notion of ''Schelling meshes'' which extends the Schelling concept to 3D meshes as a whole themselves. We collect Schelling-based data for meshes where participants are given a group of shapes and asked to choose those with the aim of matching with what they expect others to choose. We analyze the data by computing the Schelling frequency of each shape and characterizing the qualitative features that make a shape ''Schelling''. We show that the Schelling frequencies can be learned and demonstrate Schelling-guided shape applications.
Stylistic Locomotion Modeling with Conditional Variational Autoencoder
(The Eurographics Association, 2019) Du, Han; Herrmann, Erik; Sprenger, Janis; Cheema, Noshaba; hosseini, somayeh; Fischer, Klaus; Slusallek, Philipp; Cignoni, Paolo and Miguel, Eder
We propose a novel approach to create generative models for distinctive stylistic locomotion synthesis. The approach is inspired by the observation that human styles can be easily distinguished from a few examples. However, learning a generative model for natural human motions which display huge amounts of variations and randomness would require a lot of training data. Furthermore, it would require considerable efforts to create such a large motion database for each style. We propose a generative model to combine the large variation in a neutral motion database and style information from a limited number of examples. We formulate the stylistic motion modeling task as a conditional distribution learning problem. Style transfer is implicitly applied during the model learning process. A conditional variational autoencoder (CVAE) is applied to learn the distribution and stylistic examples are used as constraints. We demonstrate that our approach can generate any number of natural-looking human motions with a similar style to the target given a few style examples and a neutral motion database.
A Validation Tool For Improving Semantic Segmentation of Complex Natural Structures
(The Eurographics Association, 2019) Pavoni, Gaia; Corsini, Massimiliano; Palma, Marco; Scopigno, Roberto; Cignoni, Paolo and Miguel, Eder
The automatic recognition of natural structures is a challenging task in the supervised learning field. Complex morphologies are difficult to detect both from the networks, that may suffer from generalization issues, and from human operators, affecting the consistency of training datasets. The task of manual annotating biological structures is not comparable to a generic task of detecting an object (a car, a cat, or a flower) within an image. Biological structures are more similar to textures, and specimen borders exhibit intricate shapes. In this specific context, manual labelling is very sensitive to human error. The interactive validation of the predictions is a valuable resource to improve the network performance and address the inaccuracy caused by the lack of annotation consistency of human operators reported in literature. The proposed tool, inspired by the Yes/No Answer paradigm, integrates the semantic segmentation results coming from a CNN with the previous human labeling, allowing a more accurate annotation of thousands of instances in a short time. At the end of the validation, it is possible to obtain corrected statistics or export the integrated dataset and re-train the network.

Browse

Browsing EG 2019 - Short Papers by Subject "Computing methodologies"

Results Per Page

Sort Options