12 results
Search Results
Now showing 1 - 10 of 12
Item NURBS-based Inverse Reflector Design(The Eurographics Association, 2008) Anson, Oscar; Seron, Francisco J.; Gutierrez, Diego; Luis Matey and Juan Carlos TorresCommonly used direct rendering techniques simulate light transport for a complete scene, specified in terms of light sources, geometry, materials, participating media, etc. On the other hand, inverse rendering problems take as input a desired light distribution and try to find the unknown parts of the scene needed to get such light field. The latter kind, where inverse reflector design is included, is traditionally solved by simulation optimization methods, due to the high complexity of the inverse problem. In this paper we present an inverse reflector design method which handles surfaces as NURBS and simulates accurately the light transport by means of a modified photon mapping algorithm. The proposed method is based on an optimization method, called pattern search, in order to compute the reflector needed to generate a target near light field. Some assumptions are determined in order to reduce the complexity of the problem, such as a rotationally symmetric reflector or its perfectly specular reflective behavior. The optimization method specifies the reflector shape by handling a NURBS curve as a generatrix, sequentially modifying the position and weights of its control points in order to obtain the reflector solution. Areas of applications of inverse reflector design span from architectural lighting design to car headlamps designItem Bidirectional Clustering for Scalable VPL-based Global Illumination(The Eurographics Association, 2015) Jarabo, Adrian; Buisan, Raul; Gutierrez, Diego; Mateu Sbert and Jorge Lopez-MorenoVirtual Point Lights (VPL) methods approximate global illumination (GI) in a scene by using a large number of virtual lights modeling the reflected radiance of a surface. These methods are efficient, and allow computing noise-free images significantly faster that other methods. However, they scale linearly with the number of virtual lights and with the number of pixels to be rendered. Previous approaches improve the scalability of the method by hierarchically evaluating the virtual lights, allowing sublinear performance with respect the lights being evaluated. In this work, we introduce a novel bidirectional clustering approach, by hierarchically evaluating both the virtual lights and the shading points. This allows reusing radiance evaluation between pixels, and obtaining sublinear costs with respect to both lights and camera samples. We demonstrate significantly better performance than state-of-the-art VPL clustering methods with several examples, including high-resolution images, distributed effects, and rendering of light fields.Item Low Cost Decomposition of Direct and Global Illumination in Real Scenes(The Eurographics Association, 2015) Garces, Elena; Martin, Fernando; Gutierrez, Diego; Mateu Sbert and Jorge Lopez-MorenoRecent advances in the field of computational light transport have made it possible to solve previously unsolvable problems thanks to incorporating new devices and techniques. One of these problems is the decomposition of the illumination into its local and global components in real scenes. Previous work has managed to perform such a decomposition by projecting several light patterns on a target scene and processing its captures. In this work we build on that approach and propose two novel contributions: first, a new interpolation method, which allows the decomposition of the light components from a single capture of the projected scene into the native resolution, without requiring down-sampling; second, we propose an implementation of the algorithm for a mobile platform.Item Compressive High Speed Video Acquisition(The Eurographics Association, 2015) Serrano, Ana; Gutierrez, Diego; Masia, Belen; Mateu Sbert and Jorge Lopez-MorenoTraditional video capture is limited by the trade-off between spatial and temporal resolution. When capturing videos of high temporal resolution, the spatial resolutions decreases due to bandwidth limitations in the capture system. Achieving both high spatial and temporal resolution is only possible with highly specialized and very expensive hardware; although the bandwidth is higher, the same basic trade-off remains. In this paper, we make use of a single-shot, high-speed video capture system, in order to overcome this limitation. It is based on compressive sensing, and relies on dictionary learning for sparse video representation. This allows capturing a video sequence by coding the temporal information in a single frame, and then reconstructing the full video sequence from this single coded image. We perform an in-depth analysis of the parameters of influence in the system, providing insights for future developments of similar systems.Item Understanding Exposure for Reverse Tone Mapping(The Eurographics Association, 2008) Martin, Miguel; Fleming, Roland; Sorkine, Olga; Gutierrez, Diego; Luis Matey and Juan Carlos TorresHigh dynamic range (HDR) displays are capable of providing a rich visual experience by boosting both luminance and contrast beyond what conventional displays can offer.We envision that HDR capture and display hardware will soon reach the mass market and become mainstream in most fields, from entertainment to scientific visualization. This will necessarily lead to an extensive redesign of the imaging pipeline. However, a vast amount of legacy content is available, captured and stored using the traditional, low dynamic range (LDR) pipeline. The immediate question that arises is: will our current LDR digital material be properly visualized on an HDR display? The answer to this question involves the process known as reverse tone mapping (the expansion of luminance and contrast to match those of the HDR display) for which no definite solution exists. This paper studies the specific problem of reverse tone mapping for imperfect legacy still images, where some regions are under- or overexposed. First, we show the results of a psychophysical study compared with first-order image statistics, in an attempt to gain some understanding in what makes an image be perceived as incorrectly exposed; second, we propose a methodology to evaluate existing reverse tone mapping algorithms in the case of imperfect legacy content.Item Image-based Participating Media(The Eurographics Association, 2008) Lopez-Moreno, Jorge; Cabanes, Angel; Gutierrez, Diego; Luis Matey and Juan Carlos TorresLight transport inside participating media, like fog or water, involves complex interaction phenomena, which make traditional 3D rendering approaches challenging and computationally expensive. To circumvent this, we propose an image-based method which adds perceptually plausible participating media effects to a single, clean high dynamic range image. We impose no prior requirements on the input image, and show that the underconstrained nature of the problem (where variables like depth or reflectance properties of the objects are obviously unknown) can be overcome with relatively little unskilled user input, similar to other image-editing techniques. We additionally validate the visual correctness of the results by means of psychophysical tests.Item Analysis of Coded Apertures for Defocus Deblurring of HDR Images(The Eurographics Association, 2012) Garcia, Luis; Presa, Lara; Gutierrez, Diego; Masia, Belen; Isabel Navazo and Gustavo PatowIn recent years, research on computational photography has reached important advances in the field of coded apertures for defocus deblurring. These advances are known to perform well for low dynamic range images (LDR), but nothing is written about the extension of these techniques to high dynamic range imaging (HDR). In this paper, we focus on the analysis of how existing coded apertures techniques perform in defocus deblurring of HDR images. We present and analyse three different methods for recovering focused HDR radiances from an input of blurred LDR exposures and from a single blurred HDR radiance, and compare them in terms of the quality of their results, given by the perceptual metric HDR-VDP2. Our research includes the analysis of the employment of different statistical deconvolution priors, made both from HDR and LDR images, performing synthetic experiments as well as real ones.Item Where are the Lights? Measuring the Accuracy of Human Vision(The Eurographics Association, 2009) Lopez-Moreno, Jorge; Sangorrin, Francisco; Latorre, Pedro; Gutierrez, Diego; Carlos Andujar and Javier LluchIn real life, light sources are frequently not present in our view field. However human vision is able to infer the illumination just by observing its effect on visible objects (serving as lightprobes) or, inverting the idea, it is able to spot an object which is incoherently lit in a composition. These lightprobes have been used by computer algorithms in the same manner to detect lights, mimicking the human visual system (HVS). It has been proved that the presence of shadows or highlights in the lightprobe affects the accuracy of HVS, although its degree of influence remains unbeknownst until now. The present work performs a psychophysical analysis which aims to provide accurate data for light detection, perception-oriented rendering, image compositing and augmented reality.Item Improving Depth Estimation Using Superpixels(The Eurographics Association, 2014) Cambra, Ana B.; Muñoz, Adolfo; Murillo, Ana C.; Guerrero, José J.; Gutierrez, Diego; Adolfo Munoz and Pere-Pau VazquezThis work is focused on assigning a depth label to each pixel in the image. We consider off-the-shelf algorithms that provide depth information from multiple views or depth information directly obtained from RGB-d sensors. Both of them are common scenarios of a well studied problem where many times we get incomplete depth information. Then, user interaction becomes necessary to finish, improve or correct the solution for certain applications where accurate and dense depth information for all pixels in the image is needed. This work presents our approach to improve the depth assigned to each pixel in an automated manner. Our proposed pipeline combines state-of-the art methods for image superpixel segmentation and energy minimization. Superpixel segmentation reduces complexity and provides more robustness to the labeling decisions. We study how to propagate the depth information to incomplete or inconsistent regions of the image using a Markov Random Field (MRF) energy minimization framework. We propose and evaluate an energy function and validate it together with the designed pipeline. We present a quantitative evaluation of our approach with different variations to show the improvements we can obtain. This is done using a publicly available stereo dataset that provides ground truth information. We show additional qualitatively results, with other tests cases and scenarios using different input depth information, where we also obtain significant improvements on the depth estimation compared to the initial one.Item Depth from a Single Image Through User Interaction(The Eurographics Association, 2014) Lopez, Angeles; Garces, Elena; Gutierrez, Diego; Adolfo Munoz and Pere-Pau VazquezIn this paper we present a method to obtain a depth map from a single image of a scene by exploiting both image content and user interaction. Assuming that regions with low gradients will have similar depth values, we formulate the problem as an optimization process across a graph, where pixels are considered as nodes and edges between neighbouring pixels are assigned weights based on the image gradient. Starting from a number of userdefined constraints, depth values are propagated between highly connected nodes i.e. with small gradients. Such constraints include, for example, depth equalities and inequalities between pairs of pixels, and may include some information about perspective. This framework provides a depth map of the scene, which is useful for a number of applications.