LiDAR-captured point clouds are often considered the gold standard in active 3D reconstruction. While their accuracy is exceptional in flat regions, the capturing is susceptible to missing small geometric structures, thin edges, and structures exhibiting challenging surface properties. Alternatively, capturing multiple photos of the scene and applying 3D photogrammetry can infer these details as they often represent feature-rich regions. However, the accuracy of LiDAR for featureless regions is rarely reached. Therefore, we suggest combining the strengths of LiDAR and camera-based capture by introducing SurfFill: a Gaussian surfel-based LiDAR completion scheme. We analyze LiDAR capturings and attribute LiDAR beam divergence as a main factor for artifacts, manifesting mostly at thin structures and edges. We use this insight to introduce an ambiguity heuristic for completed scans by evaluating the change in density in the point cloud. This allows us to identify points close to missed areas, which we can then use to grow additional points from to complete the scan. For this point growing, we employ Gaussian surfels and focus optimization and densification on these ambiguous areas. Finally, Gaussian primitives of the reconstruction in ambiguous areas are extracted and sampled for points to complete the point cloud. To address the challenges of large-scale reconstruction, we extend this pipeline with a divide-and-conquer scheme for building-sized point cloud completion. We evaluate on the task of LiDAR point cloud completion of synthetic and real-world scenes and find that our method outperforms previous reconstruction methods.

Abstract 3D Gaussian Splatting (3DGS) [KKLD23] has transformed novel‐view synthesis from RGB images, yet remains restricted to the visible spectrum. Many applications, including agricultural monitoring, rely on multi‐spectral imaging, where spectral camera alignment and scalability pose major challenges. We present MS‐Splatting—a multi‐spectral 3DGS framework enabling unified multi‐view consistent reconstruction and rendering across both visible and invisible spectra. Our key component is a neural color representation that encodes per‐primitive features shared across spectral bands, decoded through a shallow multi‐layer perceptron into spectrum‐specific radiance. By leveraging inter‐band correlations, this formulation enhances detail while reducing memory consumption compared to independent band modeling via per‐channel modeling with spherical harmonics. Our method enables accurate parallax‐free novel‐view vegetation index rendering for plant monitoring and enhances RGB novel view synthesis quality by exploiting details revealed through multi‐spectral bands. Our evaluation demonstrates that MS‐Splatting exceeds the current leading methods in both categories. In addition, we introduce a multi‐spectral dataset from aerial captures covering outdoor environments, specifically designed for evaluating these applications. We will release our code and dataset to facilitate further research. The project page is located at: https://meyerls.github.io/ms_splatting

Utilizing the complementary strengths of wavelength-specific range or depth sensors is crucial for robust computer-assisted tasks such as autonomous driving. Despite this, there is still little research done at the intersection of optical depth sensors and radars operating close range, where the target is decimeters away from the sensors. Together with a growing interest in high-resolution imaging radars operating in the near field, the question arises how these sensors behave in comparison to their traditional optical counterparts. In this work, we take on the unique challenge of jointly characterizing depth imagers from both, the optical and radio-frequency domain using a multimodal spatial calibration. We collect data from four depth imagers, with three optical sensors of varying operation principle and an imaging radar. We provide a comprehensive evaluation of their depth measurements with respect to distinct object materials, geometries, and object-to-sensor distances. Specifically, we reveal scattering effects of partially transmissive materials and investigate the response of radio-frequency signals. All object measurements are made public in form of a multimodal dataset, called MAROON, which can be accessed at: https://vwirth.github.io/maroon .

Abstract Reconstructing outdoor environments from “inside-out” captures, where a camera moves within a restricted area but looks outward, remains challenging due to the presence of both well-textured nearby regions and low-detail distant backgrounds. We introduce a two-stage Gaussian Splatting framework that explicitly separates and optimizes these regions, yielding higher-fidelity novel view synthesis and allowing the replacement of the distant part with a high-quality, inpainted environment map to speedup the rendering process. In stage one, background primitives are initialized within a spherical shell and optimized using a loss that combines a background-only photometric term with two geometric regularizers: one constraining Gaussians to remain inside the shell, and another one aligning them with local tangential planes. In stage two, foreground Gaussians are initialized from a Structure-from-Motion reconstruction, added and refined using the standard rendering loss, while the background set remains fixed but contributes to the final image formation. Background Gaussians can be rendered to an object-free environment map that is inpainted to fill missing parts and can replace the Gaussian-based background for faster rendering. Experiments on diverse outdoor datasets show that our method reduces background artifacts and improves perceptual quality of novel view renderings compared to state-of-the-art baselines, including the removal of floaters in the navigation region.

We introduce a new approach for reconstruction and novel view synthesis of unbounded real-world scenes. In contrast to previous methods using either volumetric fields, grid-based models, or discrete point cloud proxies, we pro-pose a hybrid scene representation, which implicitly encodes the geometry in a continuous octree-based probability field and view-dependent appearance in a multi-resolution hash grid. This allows for extraction of arbitrary explicit point clouds, which can be rendered using rasterization. In doing so, we combine the benefits of both worlds and retain favorable behavior during optimization: Our novel implicit point cloud representation and differentiable bilinear rasterizer enable fast rendering while preserving the fine geometric detail captured by volumetric neural fields. Furthermore, this representation does not depend on priors like structure-from-motion point clouds. Our method achieves state-of-the-art image quality on common benchmarks. Furthermore, we achieve fast inference at interactive frame rates, and can convert our trained model into a large, explicit point cloud to further enhance performance.

We introduce FruitNeRF++, a novel fruit-counting approach that combines contrastive learning with neural radiance fields to count fruits from unstructured input photographs of orchards. Our work is based on FruitNeRF [6], which employs a neural semantic field combined with a fruit-specific clustering approach. The requirement for adaptation for each fruit type limits the applicability of the method, and makes it difficult to use in practice. To lift this limitation, we design a shape-agnostic multi-fruit counting framework, that complements the RGB and semantic data with instance masks predicted by a vision foundation model. The masks are used to encode the identity of each fruit as instance embeddings into a neural instance field. By volumetrically sampling the neural fields, we extract a point cloud embedded with the instance features, which can be clustered in a fruit-agnostic manner to obtain the fruit count. We evaluate our approach using a synthetic dataset containing apples, plums, lemons, pears, peaches, and mangoes, as well as a real-world benchmark apple dataset. Our results demonstrate that FruitNeRF++ is easier to control and compares favorably to other state-of-the-art methods.

Abstract 3D Gaussian Splats (3DGS) have proven a versatile rendering primitive, both for inverse rendering as well as real‐time exploration of scenes. In these applications, coherence across camera frames and multiple views is crucial, be it for robust convergence of a scene reconstruction or for artifact‐free fly‐throughs. Recent work started mitigating artifacts that break multi‐view coherence, including popping artifacts due to inconsistent transparency sorting and perspective‐correct outlines of (2D) splats. At the same time, real‐time requirements forced such implementations to accept compromises in how transparency of large assemblies of 3D Gaussians is resolved, in turn breaking coherence in other ways. In our work, we aim at achieving maximum coherence, by rendering fully perspective‐correct 3D Gaussians while using a high‐quality approximation of accurate blending, hybrid transparency, on a per‐pixel level, in order to retain real‐time frame rates. Our fast and perspectively accurate approach for evaluation of 3D Gaussians does not require matrix inversions, thereby ensuring numerical stability and eliminating the need for special handling of degenerate splats, and the hybrid transparency formulation for blending maintains similar quality as fully resolved per‐pixel transparencies at a fraction of the rendering costs. We further show that each of these two components can be independently integrated into Gaussian splatting systems. In combination, they achieve up to 2× higher frame rates, 2× faster optimization, and equal or better image quality with fewer rendering artifacts compared to traditional 3DGS on common benchmarks.

Recent advancements in point cloud compression have primarily emphasized geometry compression while comparatively fewer efforts have been dedicated to attribute compression. This study introduces an end-to-end learned dynamic lossy attribute coding approach, utilizing an efficient high-dimensional convolution to capture extensive inter-point dependencies. This enables the efficient projection of attribute features into latent variables. Subsequently, we employ a context model that leverage previous latent space in conjunction with an auto-regressive context model for encoding the latent tensor into a bitstream. Evaluation of our method on widely utilized point cloud datasets from the MPEG and Microsoft demonstrates its superior performance compared to the core attribute compression module Region-Adaptive Hierarchical Transform method from MPEG Geometry Point Cloud Compression with $38.1 \%$ Bjontegaard Delta-rate saving in average while ensuring a low-complexity encoding/decoding.

Despite an emerging interest in MIMO radar, the utilization of its complementary strengths in combination with optical depth sensors has so far been limited to far-field applications, due to the challenges that arise from mutual sensor calibration in the near field. In fact, most related approaches in the autonomous industry propose target-based calibration methods using corner reflectors that have proven to be unsuitable for the near field. In contrast, we propose a novel, joint calibration approach for optical RGB-D sensors and MIMO radars that is designed to operate in the radar’s near-field range, within decimeters from the sensors. Our pipeline consists of a bespoke calibration target, allowing for automatic target detection and localization, followed by the spatial calibration of the two sensor coordinate systems through target registration. We validate our approach using two different depth sensing technologies from the optical domain. The experiments show the efficiency and accuracy of our calibration for various target displacements, as well as its robustness of our localization in terms of signal ambiguities.

We introduce Physically Enhanced Gaussian Splatting Simulation System (PEGASUS) for 6DoF object pose dataset generation, a versatile dataset generator based on 3D Gaussian Splatting. Environment and object representations can be easily obtained using commodity cameras to reconstruct with Gaussian Splatting. PEGASUS allows the composition of new scenes by merging the respective underlying Gaussian Splatting point cloud of an environment with one or multiple objects. Leveraging a physics engine enables the simulation of natural object placement within a scene through interaction between meshes extracted for the objects and the environment. Consequently, an extensive amount of new scenes - static or dynamic - can be created by combining different environments and objects. By rendering scenes from various perspectives, diverse data points such as RGB images, depth maps, semantic masks, and 6DoF object poses can be extracted. Our study demonstrates that training on data generated by PEGASUS enables pose estimation networks to successfully transfer from synthetic data to real-world data. Moreover, we introduce the Ramen dataset, comprising 30 Japanese cup noodle items. This dataset includes spherical scans that capture images from both the object hemisphere and the Gaussian Splatting reconstruction, making them compatible with PEGASUS.

We introduce FruitNeRF, a unified novel fruit counting framework that leverages state-of-the-art view synthesis methods to count any fruit type directly in 3D. Our framework takes an unordered set of posed images captured by a monocular camera and segments fruit in each image. To make our system independent of the fruit type, we employ a foundation model that generates binary segmentation masks for any fruit. Utilizing both modalities, RGB and semantic, we train a semantic neural radiance field. Through uniform volume sampling of the implicit Fruit Field, we obtain fruit-only point clouds. By applying cascaded clustering on the extracted point cloud, our approach achieves precise fruit count. The use of neural radiance fields provides significant advantages over conventional methods such as object tracking or optical flow, as the counting itself is lifted into 3D. Our method prevents double counting fruit and avoids counting irrelevant fruit. We evaluate our methodology using both real-world and synthetic datasets. The real-world dataset consists of three apple trees with manually counted ground truths, a benchmark apple dataset with one row and ground truth fruit location, while the synthetic dataset comprises various fruit types including apple, plum, lemon, pear, peach, and mango. Additionally, we assess the performance of fruit counting using the foundation model compared to a U-Net.

Abstract Point‐based radiance field rendering has demonstrated impressive results for novel view synthesis, offering a compelling blend of rendering quality and computational efficiency. However, also latest approaches in this domain are not without their shortcomings. 3D Gaussian Splatting [KKLD23] struggles when tasked with rendering highly detailed scenes, due to blurring and cloudy artifacts. On the other hand, ADOP [RFS22] can accommodate crisper images, but the neural reconstruction network decreases performance, it grapples with temporal instability and it is unable to effectively address large gaps in the point cloud. In this paper, we present TRIPS (Trilinear Point Splatting), an approach that combines ideas from both Gaussian Splatting and ADOP. The fundamental concept behind our novel technique involves rasterizing points into a screen‐space image pyramid, with the selection of the pyramid layer determined by the projected point size. This approach allows rendering arbitrarily large points using a single trilinear write. A lightweight neural network is then used to reconstruct a hole‐free image including detail beyond splat resolution. Importantly, our render pipeline is entirely differentiable, allowing for automatic optimization of both point sizes and positions. Our evaluation demonstrate that TRIPS surpasses existing state‐of‐the‐art methods in terms of rendering quality while maintaining a real‐time frame rate of 60 frames per second on readily available hardware. This performance extends to challenging scenarios, such as scenes featuring intricate geometry, expansive landscapes, and auto‐exposed footage. The project page is located at: https://lfranke.github.io/trips

Existing real-time RGB-D reconstruction approaches, like Kinect Fusion, lack real-time photo-realistic visualization. This is due to noisy, oversmoothed or incomplete geometry and blurry textures which are fused from imperfect depth maps and camera poses. Recent neural rendering methods can overcome many of such artifacts but are mostly optimized for offline usage, hindering the integration into a live reconstruction pipeline.

In the last few years, deep neural networks opened the doors for big advances in novel view synthesis. Many of these approaches are based on a (coarse) proxy geometry obtained by structure from motion algorithms. Small deficiencies in this proxy can be fixed by neural rendering, but larger holes or missing parts, as they commonly appear for thin structures or for glossy regions, still lead to distracting artifacts and temporal instability. In this paper, we present a novel neural-rendering-based approach to detect and fix such deficiencies. As a proxy, we use a point cloud, which allows us to easily remove outlier geometry and to fill in missing geometry without complicated topological operations. Keys to our approach are (i) a differentiable, blending point-based renderer that can blend out redundant points, as well as (ii) the concept of Visual Error Tomography (VET), which allows us to lift 2D error maps to identify 3D-regions lacking geometry and to spawn novel points accordingly. Furthermore, (iii) by adding points as nested environment maps, our approach allows us to generate high-quality renderings of the surroundings in the same pipeline. In our results, we show that our approach can improve the quality of a point cloud obtained by structure from motion and thus increase novel view synthesis quality significantly. In contrast to point growing techniques, the approach can also fix large-scale holes and missing thin structures effectively. Rendering quality outperforms state-of-the-art methods and temporal stability is significantly improved, while rendering is possible at real-time frame rates.

Novel-view synthesis is an ill-posed problem in that it requires inference of previously unseen information. Recently, reviving the traditional field of image-based rendering, neural methods proved particularly suitable for this interpolation/extrapolation task; however, they often require a-priori scene-completeness or costly preprocessing steps and generally suffer from long (scene-specific) training times. Our work draws from recent progress in neural spatio-temporal supersampling to enhance a state-of-the-art neural renderer’s ability to infer novel-view information at inference time. We adapt a supersampling architecture [Xiao et al. 2020], which resamples previously rendered frames, to instead recombine nearby camera images in a multi-view dataset. These input frames are warped into a joint target frame, guided by the most recent (point-based) scene representation, followed by neural interpolation. The resulting architecture gains sufficient robustness to significantly improve transferability to previously unseen datasets. In particular, this enables novel applications for neural rendering where dynamically streamed content is directly incorporated in a (neural) image-based reconstruction of a scene. As we will show, our method reaches state-of-the-art performance when compared to previous works that rely on static and sufficiently densely sampled scenes; in addition, we demonstrate our system’s particular suitability for dynamically streamed content, where our approach is able to produce high-fidelity novel-view synthesis even with significantly fewer available frames than competing neural methods.

In this paper we present ADOP, a novel point-based, differentiable neural rendering pipeline. Like other neural renderers, our system takes as input calibrated camera images and a proxy geometry of the scene, in our case a point cloud. To generate a novel view, the point cloud is rasterized with learned feature vectors as colors and a deep neural network fills the remaining holes and shades each output pixel. The rasterizer renders points as one-pixel splats, which makes it very fast and allows us to compute gradients with respect to all relevant input parameters efficiently. Furthermore, our pipeline contains a fully differentiable physically-based photometric camera model, including exposure, white balance, and a camera response function. Following the idea of inverse rendering, we use our renderer to refine its input in order to reduce inconsistencies and optimize the quality of its output. In particular, we can optimize structural parameters like the camera pose, lens distortions, point positions and features, and a neural environment map, but also photometric parameters like camera response function, vignetting, and per-image exposure and white balance. Because our pipeline includes photometric parameters, e.g. exposure and camera response function, our system can smoothly handle input images with varying exposure and white balance, and generates high-dynamic range output. We show that due to the improved input, we can achieve high render quality, also for difficult input, e.g. with imperfect camera calibrations, inaccurate proxy geometry, or varying exposure. As a result, a simpler and thus faster deep neural network is sufficient for reconstruction. In combination with the fast point rasterization, ADOP achieves real-time rendering rates even for models with well over 100M points. https://github.com/darglein/ADOP

Projection mapping with inexpensive hardware often suffers from calibration errors that lead to visually compromised results. In this paper, we classify common errors that lead to typical visual artifacts. Based on this classification, we present the first content-aware brightness solver. It is tailored for high GPU performance, yet efficiently hides the most common calibration artifacts. Moreover, it is specifically designed to handle both single and larger networked projection mapping setups with minimal latency.

We present a pop-up projection painting system that projects onto an unknown three-dimensional surface, while the user creates the projection content on the fly. The digital paint is projected immediately and follows the object if it is moved. If unexplored surface areas are thereby exposed, an automated trigger system issues new depth recordings that expand and refine the surface estimate. By intertwining scanning and projection painting we scan the exposed surface at the appropriate time and only if needed. Like image-based rendering, multiple automatically recorded depth maps are fused in screen space to synthesize novel views of the object, making projection poses independent from the scan positions. Since the user’s digital paint is also stored in images, we eliminate the need to reconstruct and parametrize a single full mesh, which makes geometry and color updates simple and fast.

Abstract Rendering in real time for virtual reality headsets with high user immersion is challenging due to strict framerate constraints as well as due to a low tolerance for artefacts. Eye tracking‐based foveated rendering presents an opportunity to strongly increase performance without loss of perceived visual quality. To this end, we propose a novel foveated rendering method for virtual reality headsets with integrated eye tracking hardware. Our method comprises recycling pixels in the periphery by spatio‐temporally reprojecting them from previous frames. Artefacts and disocclusions caused by this reprojection are detected and re‐evaluated according to a confidence value that is determined by a newly introduced formalized perception‐based metric, referred to as confidence function. The foveal region, as well as areas with low confidence values, are redrawn efficiently, as the confidence value allows for the delicate regulation of hierarchical geometry and pixel culling. Hence, the average primitive processing and shading costs are lowered dramatically. Evaluated against regular rendering as well as established foveated rendering methods, our approach shows increased performance in both cases. Furthermore, our method is not restricted to static scenes and provides an acceleration structure for post‐processing passes.

With the ubiquity of social networks, images have become crucial in todays exchange of information. Most of these images are taken by smartphones. For forensic approaches relying on fixed image formation pipelines, the capabilities of smartphones using computational photography pose new challenges. But these new capabilities also offer opportunities for forensic analysis. A growing amount of commodity devices are able to capture 3-D information using various technologies such as stereo imaging or structured light. Modern smartphones commonly save such 3-D information as depth maps alongside regular images.In this work, we propose to use characteristic artifacts of depth reconstruction algorithms as trace for forensic analysis. The proposed method is able to infer the source algorithm of stereo reconstructions with an accuracy of up to 97%. We further demonstrate the applicability of the method to collected smartphone data. It is able to discriminate patches from different sources with an AUC of up to 0.88 and can be used for splicing localization in depth maps.

Projection mapping augments a real-world object's appearance by projecting digital content on its surface. However, a remaining obstacle to immersive projection mapping is the limitation to white Lambertian surfaces and uniform neutral environment light, if any. Violating one of these assumptions results in a discernible difference between the source material and the appearance of the projected content. For example, some colors may not be visible due to intense environment lighting or pronounced surface colors. We present a system that actively subdues many of those real-world influences, especially environment lighting. Our system supports dynamic (i.e., movable) target objects as well as changing lighting conditions while requiring no prior color calibration of the projector nor any precomputed environment probing. We automatically and continuously estimate these influences during runtime in a real-time feedback-loop and adjust the projected colors accordingly.

Multi-view Stereo (MVS) is a common solution in photogrammetry applications for the dense reconstruction of a static scene from images. The static scene assumption, however, limits the general applicability of MVS algorithms, as many day-to-day scenes undergo non-rigid motion, e.g., clothes, faces, or human bodies. In this paper, we open up a new challenging direction: Dense 3D reconstruction of scenes with non-rigid changes observed from a small number of images sparsely captured from different views with a single monocular camera, which we call non-rigid multi-view stereo (NRMVS) problem. We formulate this problem as a joint optimization of deformation and depth estimation, using deformation graphs as the underlying representation. We propose a new sparse 3D to 2D matching technique with a dense patch-match evaluation scheme to estimate the most plausible deformation field satisfying depth and photometric consistency. We show that a dense reconstruction of a scene with non-rigid changes from a few images is possible, and demonstrate that our method can be used to interpolate novel deformed scenes from various combinations of deformation estimates derived from the sparse views.

Photogrammetry Studios are a common setup to acquire high-quality 3D geometry from different kinds of real-world objects, humans, etc. In a photo studio like setup, 50 -200 DSLR cameras are used with object-specific illumination to simultaneously capture images that are processed by algorithms that automatically estimate the camera parameters and detailed geometry. These steps are automated in established pipelines to a large extent and do not require much user input. However, the post-processing typically involves a manual estimation of surface reflectance parameters by an artist, who paints textures to allow for photorealistic rendering. While professional light stages facilitate this process in an automated way, these setups are very expensive and require accurately calibrated light sources and cameras. In our work, we present a new formulation along with a practical solution to reduce these constraints to photo studio like setups by jointly reconstructing the geometric configuration of the lights along with spatially varying surface reflectance properties and its diffuse albedo. In the presented synthetic as well as real-world experiments, we analyze the effect of different optimization objectives and show that our method is able to provide photorealistic reconstruction results with an RMSE of ≈ 1 - 3% on real data.

The goal of blind image forensics is to determine authenticity and origin of an image without using an explicitly embedded security scheme. Most existing forensic methods can roughly be grouped into statistical and physics-based approaches. Statistical methods can oftentimes be fully automated, and achieve impressive results on current state-of-the-art benchmarks. Physics-based methods explain image inconsistencies using an analytic model, and are more robust to common image processing operations such as resizing or recompression. In this work, we propose a physics-based forensic descriptor to characterize 2-D lighting environments of objects. The key idea is that the integral over a gradient field of an object indicates the direction of incident light in the image plane. In contrast to prior 2-D lighting methods, the proposed method is remarkably robust to changes in object color and variations in user input, as it operates on the whole object area instead of object contours. Furthermore, we show that the proposed method is unaffected by image resizing or compression, which makes it possible to analyze images that are impossible to analyze with current state-of-the-art statistical methods.

Common VR headsets require lenses that increase the field of view and allow the user to focus the display. In most systems, thick lenses are used that generate strong pincushion distortions. To account for this, the content is warped by a corresponding barrel distortion before displaying, resulting in undistorted images for the viewer. This approach assumes that the eye is exactly positioned, typically on the optical axis of the lens. However, in real systems the eye's location deviates from this optimal position even at rest; moreover, the pupil - and thus the optical center of the eye - moves by several millimeters when the user looks around. Thus, eye movement results in additional distortion, which is ignored in current VR headsets. Also in literature on head-mounted displays, the effect is most often not considered, or at least badly documented. The contribution of this paper are experiments that emphasize the importance of this mostly ignored effect. To this end, we have built a simple setup with a camera at variable eye positions in a standard VR headset, that allows us to directly measure the variation of distortion during eye movement. Our experiments show that distortion varies by several dozens of pixels within the full range of eye movements, which emphasizes that the effect is definitely significant. We also demonstrate how in a headset with built-in eye tracker, the knowledge of the eye position can be used to achieve a view-dependent lens correction with only minimal additional effort at run-time.

Abstract Monte‐Carlo path tracing techniques can generate stunning visualizations of medical volumetric data. In a clinical context, such renderings turned out to be valuable for communication, education, and diagnosis. Because a large number of computationally expensive lighting samples is required to converge to a smooth result, progressive rendering is the only option for interactive settings: Low‐sampled, noisy images are shown while the user explores the data, and as soon as the camera is at rest the view is progressively refined. During interaction, the visual quality is low, which strongly impedes the user's experience. Even worse, when a data set is explored in virtual reality, the camera is never at rest, leading to constantly low image quality and strong flickering. In this work we present an approach to bring volumetric Monte‐Carlo path tracing to the interactive domain by reusing samples over time. To this end, we transfer the idea of temporal antialiasing from surface rendering to volume rendering. We show how to reproject volumetric ray samples even though they cannot be pinned to a particular 3D position, present an improved weighting scheme that makes longer history trails possible, and define an error accumulation method that downweights less appropriate older samples. Furthermore, we exploit reprojection information to adaptively determine the number of newly generated path tracing samples for each individual pixel. Our approach is designed for static, medical data with both volumetric and surface‐like structures. It achieves good‐quality volumetric Monte‐Carlo renderings with only little noise, and is also usable in a VR context.

Abstract We present a linear algebra framework for structured matrices and general optimization problems. The matrices and matrix operations are defined recursively to efficiently capture complex structures and enable advanced compiler optimization. In addition to common dense and sparse matrix types, we define mixed matrices, which allow every element to be of a different type. Using mixed matrices, the low‐ and high‐level structure of complex optimization problems can be encoded in a single type. This type is then analyzed at compile time by a recursive linear solver that picks the optimal algorithm for the given problem. For common computer vision problems, our system yields a speedup of 3–5 compared to other optimization frameworks. The BLAS performance is benchmarked against the MKL library. We achieve a significant speedup in block‐SPMV and block‐SPMM. This work is implemented and released open‐source as a header‐only extension to the C+ + math library Eigen.

This paper shows that motion vectors representing the true motion of an object in a scene can be exploited to improve the encoding process of computer generated video sequences. Therefore, a set of sequences is presented for which the true motion vectors of the corresponding objects were generated on a per-pixel basis during the rendering process. In addition to conventional motion estimation methods, it is proposed to exploit the computer generated motion vectors to enhance the rate-distortion performance. To this end, a motion vector mapping method including disocclusion handling is presented. It is shown that mean rate savings of 3.78% can be achieved.

The computation of partial occlusion, as required for ambient occlusion or soft shadows, provides visually important cues but is notoriously expensive. In this paper we propose a novel solution to the ambient occlusion problem, combining signed distance scene representations and machine learning. We demonstrate how to learn and apply mappings which approximate a ray traced ground truth occlusion using only a few nearby samples of a signed distance representation. As representation for our trained mappings we use small feed-forward neural networks which are fast to evaluate, allowing for real-time occlusion queries. Our ambient occlusion approximation outperforms state-of-the-art methods in both quality and performance, yielding temporally stable and smooth results. Since our training data is different from typical machine learning approaches which mostly deal with 2D/3D image data and our techniques are also applicable to other occlusion problems (e.g. soft shadows), we give an in-depth overview of our framework. Furthermore, we discuss arising artifacts and possible extensions of our approach.

Abstract The computer graphics and vision communities have dedicated long standing efforts in building computerized tools for reconstructing, tracking, and analyzing human faces based on visual input. Over the past years rapid progress has been made, which led to novel and powerful algorithms that obtain impressive results even in the very challenging case of reconstruction from a single RGB or RGB‐D camera. The range of applications is vast and steadily growing as these technologies are further improving in speed, accuracy, and ease of use. Motivated by this rapid progress, this state‐of‐the‐art report summarizes recent trends in monocular facial performance capture and discusses its applications, which range from performance‐based animation to real‐time facial reenactment. We focus our discussion on methods where the central task is to recover and track a three dimensional model of the human face using optimization‐based reconstruction algorithms. We provide an in‐depth overview of the underlying concepts of real‐world image formation, and we discuss common assumptions and simplifications that make these algorithms practical. In addition, we extensively cover the priors that are used to better constrain the under‐constrained monocular reconstruction problem, and discuss the optimization techniques that are employed to recover dense, photo‐geometric 3D face models from monocular 2D data. Finally, we discuss a variety of use cases for the reviewed algorithms in the context of motion capture, facial animation, as well as image and video editing.

We propose FaceVR , a novel image-based method that enables video teleconferencing in VR based on self-reenactment. State-of-the-art face tracking methods in the VR context are focused on the animation of rigged 3D avatars (Li et al. 2015; Olszewski et al. 2016). Although they achieve good tracking performance, the results look cartoonish and not real. In contrast to these model-based approaches, FaceVR enables VR teleconferencing using an image-based technique that results in nearly photo-realistic outputs. The key component of FaceVR is a robust algorithm to perform real-time facial motion capture of an actor who is wearing a head-mounted display (HMD), as well as a new data-driven approach for eye tracking from monocular videos. Based on reenactment of a prerecorded stereo video of the person without the HMD, FaceVR incorporates photo-realistic re-rendering in real time, thus allowing artificial modifications of face and eye appearances. For instance, we can alter facial expressions or change gaze directions in the prerecorded target video. In a live setup, we apply these newly introduced algorithmic components.

We propose HeadOn, the first real-time source-to-target reenactment approach for complete human portrait videos that enables transfer of torso and head motion, face expression, and eye gaze. Given a short RGB-D video of the target actor, we automatically construct a personalized geometry proxy that embeds a parametric head, eye, and kinematic torso model. A novel realtime reenactment algorithm employs this proxy to photo-realistically map the captured motion from the source actor to the target actor. On top of the coarse geometric proxy, we propose a video-based rendering technique that composites the modified target portrait video via view- and pose-dependent texturing, and creates photo-realistic imagery of the target actor under novel torso and head poses, facial expressions, and gaze directions. To this end, we propose a robust tracking of the face and torso of the source actor. We extensively evaluate our approach and show significant improvements in enabling much greater flexibility in creating realistic reenacted output videos.

The quality of every dynamic multi-projection mapping system is limited by the quality of the projector to tracking device calibration. Common problems with poor calibration result in noticeable artifacts for the user, such as ghosting and seams. In this work we introduce a new, fully automated calibration algorithm that is tailored to reduce these artifacts, based on consumer-grade hardware. We achieve this goal by repurposing a structured-light scanning setup. A structured-light scanner can generate 3D geometry based on a known intrinsic and extrinsic calibration of its components (projector and RGB camera). We revert this process by providing the resulting 3D model to determine the intrinsic and extrinsic parameters of our setup (including those of a variety of tracking systems). Our system matches features and solves for all parameters in a single pass while respecting the lower quality of our sensory input.

Projection-based mixed reality is an effective tool to create immersive visualizations on real-world objects. Its wide range of applications includes art installations, education, stage shows, and advertising. In this work, we enhance a multi-projector system for dynamic projection mapping by handling various physical stray-light effects: interreflection, projector black-level, and environmental light in real time for dynamic scenes. We show how all these effects can be efficiently simulated and accounted for at runtime, resulting in significantly improved projection mapping results. By adding a global optimization step, we can further increase the dynamic range of the projection.

Parametric surfaces are an essential modeling tool in computer aided design and movie production. Even though their use is well established in industry, generating ray-traced images adds significant cost in time and memory consumption. Ray tracing such surfaces is usually accomplished by subdividing the surfaces on the fly, or by conversion to a polygonal representation. However, on-the-fly subdivision is computationally very expensive, whereas polygonal meshes require large amounts of memory. This is a particular problem for parametric surfaces with displacement, where very fine tessellation is required to faithfully represent the shape. Hence, memory restrictions are the major challenge in production rendering. In this article, we present a novel solution to this problem. We propose a compression scheme for a priori Bounding Volume Hierarchies (BVHs) on parametric patches, that reduces the data required for the hierarchy by a factor of up to 48. We further propose an approximate evaluation method that does not require leaf geometry, yielding an overall reduction of memory consumption by a factor of 60 over regular BVHs on indexed face sets and by a factor of 16 over established state-of-the-art compression schemes. Alternatively, our compression can simply be applied to a standard BVH while keeping the leaf geometry, resulting in a compression rate of up to 2:1 over current methods. Although decompression generates additional costs during traversal, we can manage very complex scenes even on the memory restrictive GPU at competitive render times.

Recent publications and art performances demonstrate amazing results using projection mapping. To our knowledge, there exists no multi-projection system that can project onto non-rigid target geometries. This constrains the applicability and quality for live performances with multiple spectators. Given the cost and complexity of current systems, we present a low-cost easy-to-use markerless non-rigid face multi-projection system. It is based on a non-rigid, dense face tracker and a real-time multi-projection solver adapted to imprecise tracking, geometry and calibration. Using this novel system we produce compelling results with only consumer-grade hardware.

We present a novel approach for real-time facial reenactment of a monocular target video sequence (e.g., Youtube video). The source sequence is also a monocular video stream, captured live with a commodity webcam. Our goal is to animate the facial expressions of the target video by a source actor and re-render the manipulated output video in a photo-realistic fashion. To this end, we first address the under-constrained problem of facial identity recovery from monocular video by non-rigid model-based bundling. At run time, we track facial expressions of both source and target video using a dense photometric consistency measure. Reenactment is then achieved by fast and efficient deformation transfer between source and target. The mouth interior that best matches the re-targeted expression is retrieved from the target sequence and warped to produce an accurate fit. Finally, we convincingly re-render the synthesized target face on top of the corresponding video stream such that it seamlessly blends with the real-world illumination. We demonstrate our method in a live setup, where Youtube videos are reenacted in real time.

Abstract Graphics hardware has progressively been optimized to render more triangles with increasingly flexible shading. For highly detailed geometry, interactive applications restricted themselves to performing transforms on fixed geometry, since they could not incur the cost required to generate and transfer smooth or displaced geometry to the GPU at render time. As a result of recent advances in graphics hardware, in particular the GPU tessellation unit, complex geometry can now be generated on the fly within the GPU's rendering pipeline. This has enabled the generation and displacement of smooth parametric surfaces in real‐time applications. However, many well‐established approaches in offline rendering are not directly transferable due to the limited tessellation patterns or the parallel execution model of the tessellation stage. In this survey, we provide an overview of recent work and challenges in this topic by summarizing, discussing, and comparing methods for the rendering of smooth and highly detailed surfaces in real time.

Rendering performance is an everlasting goal of computer graphics and significant driver for advances in both, hardware architecture and algorithms. Thereby, it has become possible to apply advanced computer graphics technology even in low-cost embedded appliances, such as car instruments. Yet, to come up with an efficient implementation, developers have to put enormous efforts into hardware/problem-specific tailoring, fine-tuning, and domain exploration, which requires profound expert knowledge. If a good solution has been found, there is a high probability that it does not work as well with other architectures or even the next hardware generation. Generative DSL-based approaches could mitigate these efforts and provide for an efficient exploration of algorithmic variants and hardware-specific tuning ideas. However, in vertically organized industries, such as automotive, suppliers are reluctant to introduce these techniques as they fear loss of control, high introduction costs, and additional constraints imposed by the OEM with respect to software and tool-chain certification. Moreover, suppliers do not want to share their generic solutions with the OEM, but only concrete instances. To this end, we propose a light-weight and incremental approach for meta programming of graphics applications. Our approach relies on an existing formulation of C-like languages that is amenable to meta programming, which we extend to become a lightweight language to combine algorithmic features. Our method provides a concise notation for meta programs and generates easily sharable output in the appropriate C-style target language.

Feature-adaptive subdivision (FAS) is one of the state-of-the art real-time rendering methods for subdivision surfaces on modern GPUs. It enables efficient and accurate rendering of subdivision surfaces in many interactive applications, such as video games or authoring tools. In this paper, we present dynamic feature-adaptive subdivision (DFAS), which improves upon FAS by enabling an independent subdivision depth for every irregularity. Our subdivision kernels fill a dynamic patch buffer on-the-fly with the appropriate number of patches corresponding to the chosen level-of-detail scheme. By reducing the number of generated and processed patches, DFAS significantly improves upon the performance of static FAS.

We present a novel technique for rendering depth of field that addresses difficult overlap cases, such as close, but out-of-focus, geometry in the near-field. Such scene configurations are not managed well by state-of-the-art post-processing approaches since essential information is missing due to occlusion.

Abstract We present a novel method to adaptively apply modifications to scene data stored in GPU memory. Such modifications may include interactive painting and sculpting operations in an authoring tool, or deformations resulting from collisions between scene objects detected by a physics engine. We only allocate GPU memory for the faces affected by these modifications to store fine‐scale colour or displacement values. This requires dynamic GPU memory management in order to assign and adaptively apply edits to individual faces at runtime. We present such a memory management technique based on a scan‐operation that is efficiently parallelizable. Since our approach runs entirely on the GPU, we avoid costly CPU–GPU memory transfer and eliminate typical bandwidth limitations. This minimizes runtime overhead to under a millisecond and makes our method ideally suited to many real‐time applications such as video games and interactive authoring tools. In addition, our algorithm significantly reduces storage requirements and allows for much higher resolution content compared to traditional global texturing approaches. Our technique can be applied to various mesh representations, including Catmull–Clark subdivision surfaces, as well as standard triangle and quad meshes. In this paper, we demonstrate several scenarios for these mesh types where our algorithm enables adaptive mesh refinement, local surface deformations and interactive on‐mesh painting and sculpting.

Abstract In this paper, we introduce a novel technique for pre‐filtering multi‐layer shadow maps. The occluders in the scene are stored as variable‐length lists of fragments for each texel. We show how this representation can be filtered by progressively merging these lists. In contrast to previous pre‐filtering techniques, our method better captures the distribution of depth values, resulting in a much higher shadow quality for overlapping occluders and occluders with different depths. The pre‐filtered maps are generated and evaluated directly on the GPU, and provide efficient queries for shadow tests with arbitrary filter sizes. Accurate soft shadows are rendered in real‐time even for complex scenes and difficult setups. Our results demonstrate that our pre‐filtered maps are general and particularly scalable.

Spherical Fibonacci point sets yield nearly uniform point distributions on the unit sphere S 2 ⊂ R 3 . The forward generation of these point sets has been widely researched and is easy to implement, such that they have been used in various applications. Unfortunately, the lack of an efficient mapping from points on the unit sphere to their closest spherical Fibonacci point set neighbors rendered them impractical for a wide range of applications, especially in computer graphics. Therefore, we introduce an inverse mapping from points on the unit sphere which yields the nearest neighbor in an arbitrarily sized spherical Fibonacci point set in constant time, without requiring any precomputations or table lookups. We show how to implement this inverse mapping on GPUs while addressing arising floating point precision problems. Further, we demonstrate the use of this mapping and its variants, and show how to apply it to fast unit vector quantization. Finally, we illustrate the means by which to modify this inverse mapping for texture mapping with smooth filter kernels and showcase its use in the field of procedural modeling.

Using projection mapping enables us to bring virtual worlds into shared physical spaces. In this paper, we present a novel, adaptable and real-time projection mapping system, which supports multiple projectors and high quality rendering of dynamic content on surfaces of complex geometrical shape. Our system allows for smooth blending across multiple projectors using a new optimization framework that simulates the diffuse direct light transport of the physical world to continuously adapt the color output of each projector pixel. We present a real-time solution to this optimization problem using off-the-shelf graphics hardware, depth cameras and projectors. Our approach enables us to move projectors, depth camera or objects while maintaining the correct illumination, in realtime, without the need for markers on the object. It also allows for projectors to be removed or dynamically added, and provides compelling results with only commodity hardware.

We present a method for the real-time transfer of facial expressions from an actor in a source video to an actor in a target video, thus enabling the ad-hoc control of the facial expressions of the target actor. The novelty of our approach lies in the transfer and photorealistic re-rendering of facial deformations and detail into the target video in a way that the newly-synthesized expressions are virtually indistinguishable from a real video. To achieve this, we accurately capture the facial performances of the source and target subjects in real-time using a commodity RGB-D sensor. For each frame, we jointly fit a parametric model for identity, expression, and skin reflectance to the input color and depth data, and also reconstruct the scene lighting. For expression transfer, we compute the difference between the source and target expressions in parameter space, and modify the target parameters to match the source expressions. A major challenge is the convincing re-rendering of the synthesized target face into the corresponding video stream. This requires a careful consideration of the lighting and shading design, which both must correspond to the real-world environment. We demonstrate our method in a live setup, where we modify a video conference feed such that the facial expressions of a different person (e.g., translator) are matched in real-time.

We present a novel method to obtain fine-scale detail in 3D reconstructions generated with low-budget RGB-D cameras or other commodity scanning devices. As the depth data of these sensors is noisy, truncated signed distance fields are typically used to regularize out the noise, which unfortunately leads to over-smoothed results. In our approach, we leverage RGB data to refine these reconstructions through shading cues, as color input is typically of much higher resolution than the depth data. As a result, we obtain reconstructions with high geometric detail, far beyond the depth resolution of the camera itself. Our core contribution is shading-based refinement directly on the implicit surface representation, which is generated from globally-aligned RGB-D images. We formulate the inverse shading problem on the volumetric distance field, and present a novel objective function which jointly optimizes for fine-scale surface geometry and spatially-varying surface reflectance. In order to enable the efficient reconstruction of sub-millimeter detail, we store and process our surface using a sparse voxel hashing scheme which we augment by introducing a grid hierarchy. A tailored GPU-based Gauss-Newton solver enables us to refine large shape models to previously unseen resolution within only a few seconds.

ABSTRACT We present a novel method for the interactive markerless reconstruction of human heads using a single commodity RGB‐D sensor. Our entire reconstruction pipeline is implemented on the graphics processing unit and allows to obtain high‐quality reconstructions of the human head using an interactive and intuitive reconstruction paradigm. The core of our method is a fast graphics processing unit‐based nonlinear quasi‐Newton solver that allows us to leverage all information of the RGB‐D stream and fit a statistical head model to the observations at interactive frame rates. By jointly solving for shape, albedo and illumination parameters, we are able to reconstruct high‐quality models including illumination corrected textures. All obtained reconstructions have a common topology and can be directly used as assets for games, films and various virtual reality applications. We show motion retargeting, retexturing and relighting examples. The accuracy of the presented algorithm is evaluated by a comparison against ground truth data. Copyright © 2014 John Wiley & Sons, Ltd.

We present a combined hardware and software solution for markerless reconstruction of non-rigidly deforming physical objects with arbitrary shape in real-time . Our system uses a single self-contained stereo camera unit built from off-the-shelf components and consumer graphics hardware to generate spatio-temporally coherent 3D models at 30 Hz. A new stereo matching algorithm estimates real-time RGB-D data. We start by scanning a smooth template model of the subject as they move rigidly. This geometric surface prior avoids strong scene assumptions, such as a kinematic human skeleton or a parametric shape model. Next, a novel GPU pipeline performs non-rigid registration of live RGB-D data to the smooth template using an extended non-linear as-rigid-as-possible (ARAP) framework. High-frequency details are fused onto the final mesh using a linear deformation model. The system is an order of magnitude faster than state-of-the-art methods, while matching the quality and robustness of many offline algorithms. We show precise real-time reconstructions of diverse scenes, including: large deformations of users' heads, hands, and upper bodies; fine-scale wrinkles and folds of skin and clothing; and non-rigid interactions performed by users on flexible objects such as toys. We demonstrate how acquired models can be used for many interactive scenarios, including re-texturing, online performance capture and preview, and real-time shape and motion re-targeting.

Dedicated visualization methods are among the most important tools of modern computer-aided medical applications. Reformation methods such as Multiplanar Reformation or Curved Planar Reformation have evolved as useful tools that facilitate diagnostic and therapeutic work. In this paper, we present a novel approach that can be seen as a generalization of Multiplanar Reformation to curved surfaces. The main concept is to generate reformatted medical volumes driven by the individual anatomical geometry of a specific patient. This process generates flat views of anatomical structures that facilitate many tasks such as diagnosis, navigation and annotation. Our reformation framework is based on a non-linear as-rigid-as-possible volumetric deformation scheme that uses generic triangular surface meshes as input. To manage inevitable distortions during reformation, we introduce importance maps which allow controlling the error distribution and improving the overall visual quality in areas of elevated interest. Our method seamlessly integrates with well-established concepts such as the slice-based inspection of medical datasets and we believe it can improve the overall efficiency of many medical workflows. To demonstrate this, we additionally present an integrated visualization system and discuss several use cases that substantiate its benefits.

Online 3D reconstruction is gaining newfound interest due to the availability of real-time consumer depth cameras. The basic problem takes live overlapping depth maps as input and incrementally fuses these into a single 3D model. This is challenging particularly when real-time performance is desired without trading quality or scale. We contribute an online system for large and fine scale volumetric reconstruction based on a memory and speed efficient data structure. Our system uses a simple spatial hashing scheme that compresses space, and allows for real-time access and updates of implicit surface data, without the need for a regular or hierarchical grid data structure. Surface data is only stored densely where measurements are observed. Additionally, data can be streamed efficiently in or out of the hash table, allowing for further scalability during sensor motion. We show interactive reconstructions of a variety of scenes, reconstructing both fine-grained details and large scale environments. We illustrate how all parts of our pipeline from depth map pre-processing, camera pose estimation, depth map fusion, and surface rendering are performed at real-time rates on commodity graphics hardware. We conclude with a comparison to current state-of-the-art online systems, illustrating improved performance and reconstruction quality.

The precise modeling of vascular structures plays a key role in medical imaging applications, such as diagnosis, therapy planning and blood flow simulations. For the simulation of blood flow in particular, high-precision models are required to produce accurate results. It is thus common practice to perform extensive manual data polishing on vascular segmentations prior to simulation. This usually involves a complex tool chain which is highly impractical for clinical on-site application. To close this gap in current blood flow simulation pipelines, we present a novel technique for interactive vascular modeling which is based on implicit sweep surfaces. Our method is able to generate and correct smooth high-quality models based on geometric centerline descriptions on the fly. It supports complex vascular free-form contours and consequently allows for an accurate and fast modeling of pathological structures such as aneurysms or stenoses. We extend the concept of implicit sweep surfaces to achieve increased robustness and applicability as required in the medical field. We finally compare our method to existing techniques and provide case studies that confirm its contribution to current simulation pipelines.

Hardware tessellation is one of the latest GPU features. Triangle or quad meshes are tessellated on-the-fly, where the tessellation level is chosen adaptively in a separate shader. The hardware tessellator only generates topology; attributes such as positions or texture coordinates of the newly generated vertices are determined in a domain shader. Typical applications of hardware tessellation are view dependent tessellation of parametric surfaces and displacement mapping. Often, the attributes for the newly generated vertices are stored in textures, which requires uv unwrapping, chartification, and atlas generation of the input mesh--a process that is time consuming and often requires manual intervention. In this paper, we present an alternative representation that directly stores optimized attribute values for typical hardware tessellation patterns and simply assigns these attributes to the generated vertices at render time. Using a multilevel fitting approach, the attribute values are optimized for several resolutions. Thereby, we require no parameterization, save memory by adapting the density of the samples to the content, and avoid discontinuities by construction. Our representation is optimally suited for displacement mapping: it automatically generates seamless, view-dependent displacement mapped models. The multilevel fitting approach generates better low-resolution displacement maps than simple downfiltering. By properly blending levels, we avoid artifacts such as popping or swimming surfaces. We also show other possible applications such as signal-optimized texturing or light baking. Our representation can be evaluated in a pixel shader, resulting in signal adaptive, parameterization-free texturing, comparable to PTex or Mesh Colors. Performance evaluation shows that our representation is on par with standard texture mapping and can be updated in real time, allowing for application such as interactive sculpting.

We present a novel representation for storing sub-triangle signals, such as colors, normals, or displacements directly with the triangle mesh. Signal samples are stored as guided by hardware-tessellation patterns. Thus, we can directly render from our representation by assigning signal samples to attributes of vertices generated by the hardware tessellator.

Abstract Accurate visualizations of complex vascular structures are essential for medical applications, such as diagnosis, therapy planning and medical education. Vascular trees are usually described using centerlines, since they capture both the topology and the geometry of the vasculature in an intuitive manner. State‐of‐the‐art vessel segmentation algorithms deliver vascular outlines as free‐form contours along the centerline, since this allows capturing anatomical pathologies. However, existing methods for generating surface representations from centerlines can only cope with circular outlines. We present a novel model‐based technique that is capable of generating intersection‐free surfaces from centerlines with complex outlines. Vascular segments are described by local signed distance functions and combined using Boolean operations. An octree‐based surface generation strategy automatically computes watertight, scale‐adaptive meshes with a controllable quality. In contrast to other approaches, our method generates a reliable representation that guarantees to capture all vessels regardless of their size.

Abstract We propose a lossless, single‐rate triangle mesh topology codec tailored for fast data‐parallel GPU decompression. Our compression scheme coherently orders generalized triangle strips in memory. To unpack generalized triangle strips efficiently, we propose a novel parallel and scalable algorithm. We order vertices coherently to further improve our compression scheme. We use a variable bit‐length code for additional compression benefits, for which we propose a scalable data‐parallel decompression algorithm. For a set of standard benchmark models, we obtain (min: 3.7, med: 4.6, max: 7.6) bits per triangle. Our CUDA decompression requires only about 15% of the time it takes to render the model even with a simple shader.

In this research paper we investigate the applicability of real-time generated computer graphics for mono camera lane and vehicle detection algorithms. First we introduce a developed hardware-in-the-loop simulator and describe two solutions for the input of the synthetic images. Then the creation of the camera and environment model used for the tests is focused. Therefore first requirements hold by the lane and vehicle detection algorithms are mentioned and different fields of application are described. Finally we demonstrate the applicability on three scenarios derived from real test drives and compare the results to reality.

Abstract We present a simple algorithm for computing a high‐quality personalized avatar from a single color image and the corresponding depth map which have been captured by Microsoft's Kinect sensor. Due to the low market price of our hardware setup, 3D face scanning becomes feasible for home use. The proposed algorithm combines the advantages of robust non‐rigid registration and fitting of a morphable face model. We obtain a high‐quality reconstruction of the facial geometry and texture along with one‐to‐one correspondences with our generic face model. This representation allows for a wide range of further applications such as facial animation or manipulation. Our algorithm has proven to be very robust. Since it does not require any user interaction, even non‐expert users can easily create their own personalized avatars. Copyright © 2011 John Wiley & Sons, Ltd.

Virtual environments are typically textured by manually choosing an image to apply on each surface. This implies browsing through large sets of generic textures for each and every surface in the scene.

Abstract In this paper we analyze normal vector representations. We derive the error of the most widely used representation, namely 3D floating‐point normal vectors. Based on this analysis, we show that, in theory, the discretization error inherent to single precision floating‐point normals can be achieved by 2 50.2 uniformly distributed normals, addressable by 51 bits. We review common sphere parameterizations and show that octahedron normal vectors perform best: they are fast and stable to compute, have a controllable error, and require only 1 bit more than the theoretical optimal discretization with the same error.

Abstract Compact surface descriptions like higher‐order surfaces are popular representations for both modeling and animation. However, for fast graphics‐hardware‐assisted rendering, they usually need to be converted to triangle meshes. In this paper, we introduce a new framework for performing on‐the‐fly crack‐free adaptive tessellation of surface primitives completely on the GPU. Utilizing CUDA and its flexible memory write capabilities, we parallelize the tessellation task at the level of single surface primitives. We are hence able to derive tessellation factors, perform surface evaluation as well as generate the tessellation topology in real‐time even for large collections of primitives. We demonstrate the power of our framework by exemplarily applying it to both bicubic rational Bézier patches and PN triangles.

Abstract The goal of texture synthesis is to generate an arbitrarily large high‐quality texture from a small input sample. Generally, it is assumed that the input image is given as a flat, square piece of texture, thus it has to be carefully prepared from a picture taken under ideal conditions. Instead we would like to extract the input texture from any surface from within an arbitrary photograph. This introduces several challenges: Only parts of the photograph are covered with the texture of interest, perspective and scene geometry introduce distortions, and the texture is non‐uniformly sampled during the capture process. This breaks many of the assumptions used for synthesis. In this paper we combine a simple novel user interface with a generic per‐pixel synthesis algorithm to achieve high‐quality synthesis from a photograph. Our interface lets the user locally describe the geometry supporting the textures by combining rational Bézier patches. These are particularly well suited to describe curved surfaces under projection. Further, we extend per‐pixel synthesis to account for arbitrary texture sparsity and distortion, both in the input image and in the synthesis output. Applications range from synthesizing textures directly from photographs to high‐quality texture completion.

Abstract Recently, several real‐time soft shadow algorithms have been introduced which all compute a single shadow map and use its texels to obtain a discrete scene representation. The resulting micropatches are backprojected onto the light source and the light areas occluded by them get accumulated to estimate overall light occlusion. This approach ignores patch overlaps, however, which can lead to objectionable artifacts. In this paper, we propose to determine the visibility of the light source with a bit field where each bit tracks the visibility of a sample point on the light source. This approach not only avoids overlapping‐related artifacts but offers a solution to the important occluder fusion problem. Hence, it also becomes possible to correctly incorporate information from multiple depth maps. In addition, a new interpretation of the shadow map data is suggested which often provides superior visual results. Finally, we show how the search area for potential occluders can be reduced substantially.

We reformulate the rendering equation to alleviate the need for explicit visibility computation, thus enabling interactive global illumination on graphics hardware. This is achieved by treating visibility implicitly and propagating an additional quantity, called antiradiance , to compensate for light transmitted extraneously. Our new algorithm shifts visibility computation to simple local iterations by maintaining additional directional antiradiance information with samples in the scene. It is easy to parallelize on a GPU. By correctly treating discretization and filtering, we can compute indirect illumination in scenes with dynamic objects much faster than traditional methods. Our results show interactive update of indirect illumination with moving characters and lights.

In this paper we present a novel method for plausible real-time rendering of indirect illumination effects for diffuse and non-diffuse surfaces. The scene geometry causing indirect illumination is captured by an extended shadow map, as proposed in previous work, and secondary light sources are distributed on directly lit surfaces. One novelty is the rendering of these secondary lights' contribution by splatting in a deferred shading process, which decouples rendering time from scene complexity. An importance sampling strategy, implemented entirely on the GPU, allows efficient selection of secondary light sources. Adapting the light's splat shape to surface glossiness also allows efficient rendering of caustics. Unlike previous approaches the approximated indirect lighting does barely exhibit coarse artifacts - even under unfavorable viewing and lighting conditions. We describe an implementation on contemporary graphics hardware, show a comparison to previous approaches, and present adaptation to and results in game-typical applications.

In this paper we present "reflective shadow maps", an algorithm for interactive rendering of plausible indirect illumination. A reflective shadow map is an extension to a standard shadow map, where every pixel is considered as an indirect light source. The illumination due to these indirect lights is evaluated on-the-fly using adaptive sampling in a fragment shader. By using screen-space interpolation of the indirect lighting, we achieve interactive rates, even for complex scenes. Since we mainly work in screen space, the additional effort is largely independent of scene complexity. The resulting indirect light is approximate, but leads to plausible results and is suited for dynamic scenes. We describe an implementation on current graphics hardware and show results achieved with our approach.

Multi-dimensional transfer functions are an efficient way to visualize features in scalar volume data produced by CT or MRI scanners. However, the optimal transfer function is difficult to find in general. We present an automatic yet powerful method for the automatic setup of multi-dimensional transfer functions by adding spatial information to the histogram of a volume. Using this information we can easily classify the histogram and derive a transfer function by assigning unique colors to each class of the histogram. Each feature can be selected interactively by pointing and clicking at the corresponding class in the transfer function. In order to render the classified volume with adequate quality we propose an extension of the wellknown pre-integration technique. Furthermore, we demonstrate the flexibility of our approach by giving examples for the imaging of segmented, diffusion-tensor and multi-modal data.

Diffusion tensor imaging is a magnetic resonance imaging method which has gained increasing importance in neuroscience and especially in neurosurgery. It acquires diffusion properties represented by a symmetric 2nd order tensor for each voxel in the gathered dataset. From the medical point of view, the data is of special interest due lo different diffusion characteristics of varying brain tissue allowing conclusions about the underlying structures such as while matter tracts. An obvious way to visualize this data is to focus on the anisotropic areas using the major eigenvector for tractography and rendering lines for visualization of the simulation results. Our approach extends this technique to avoid line representation since lines lead 10 very complex illustrations and furthermore are mistakable. Instead, we generate surfaces wrapping bundles of lines. Thereby, a more intuitive representation of different tracts is achieved.

In this paper we present sequential point trees, a data structure that allows adaptive rendering of point clouds completely on the graphics processor. Sequential point trees are based on a hierarchical point representation, but the hierarchical rendering traversal is replaced by sequential processing on the graphics processor, while the CPU is available for other tasks. Smooth transition to triangle rendering for optimized performance is integrated. We describe optimizations for backface culling and texture adaptive point selection. Finally, we discuss implementation issues and show results.

We present a method for interactive rendering of large outdoor scenes. Complex polygonal plant models and whole plant populations are represented by relatively small sets of point and line primitives. This enables us to show landscapes faithfully using only a limited percentage of primitives. In addition, a hierarchical data structure allows us to smoothly reduce the geometrical representation to any desired number of primitives. The scene is hierarchically divided into local portions of geometry to achieve large reduction factors for distant regions. Additionally, the data reduction is adapted to the visual importance of geometric objects. This allows us to maintain the visual fidelity of the representation while reducing most of the geometry drastically. With our system, we are able to interactively render very complex landscapes with good visual quality.

Radiosity methods handle large scenes and complex objects using clustering techniques. To reconstruct a high quality image, usually a second very time consuming final gather pass is applied which exactly recomputes the last light transport before reaching the eye. We propose a new final gather technique which is especially suited for scenes with fine polygonal geometry. In such scenes, substantial parts of the incident illumination vary only smoothly across the surfaces and can be reconstructed on a much coarser structure. We therefore propose a final gather reconstruction based on an object-independent 3D grid. The illumination of each sender is investigated separately: If it varies smoothly across a grid cell, it is interpolated between the vertices of the grid cell, or recomputed exactly, otherwise. We further reduce the number of required samples using view-dependent optimizations. So complex objects with a very detailed structure—plants are good example here—exhibit strong masking effects, which can be exploited by our method. Finally, the estimation of penumbra screen sizes can be used to further reduce costly visibility reevaluations. Categories and Subject Descriptors (according to ACM CCS): I.3.3 [Copmuter Graphics]: Picture/Image generation I.3.7 [Copmuter Graphics]: Three-Dimensional Graphics and Realism

Shadow maps are probably the most widely used means for the generation of shadows, despite their well known aliasing problems. In this paper we introduce perspective shadow maps, which are generated in normalized device coordinate space, i.e., after perspective transformation. This results in important reduction of shadow map aliasing with almost no overhead. We correctly treat light source transformations and show how to include all objects which cast shadows in the transformed space. Perspective shadow maps can directly replace standard shadow maps for interactive hardware accelerated rendering as well as in high-quality, offline renderers.

Using well-adapted object hierarchies can support the rendering of large scenes in different ways. For instance, the quality of the lighting simulation may be improved, or the computational cost for rendering may be reduced. However the meaning of "well-adapted" depends heavily on the criterion due to which the hierarchy has been constructed. Different applications typically have different demands like low average intersection cost for a ray tracer or grouping objects with similar material properties or surface orientation for hierarchical radiosity. The authors propose an algorithm for the automatic creation of object hierarchies. The hierarchies are constructed by sequentially inserting all scene objects into the hierarchy created so far. By basing the insertion decision on a cost function defined by the user, the method can be guided to create hierarchies tailored to the desired application. The results can be improved significantly by running a global optimization on the completed hierarchy. During this optimization step, we perform a re-grouping of the objects in the hierarchy. Any ill-formed groups that were created during the initial algorithm are subject to being eliminated by our global optimization.

When a rendering algorithm has created a pixel array of radiance values the task of producing an image is not yet completed. In fact, to visualize the result the radiance values still have to be mapped to luminances, which can be reproduced by the used display. This step is performed with the help of tone reproduction operators. These tools have mainly been applied to still images, but of course they are just as necessary for walkthrough applications, in which several images are created per second. In this paper we illuminate the physiological aspects of tone reproduction for interactive applications. It is shown how tone reproduction can also be introduced into interactive radiosity viewers, where the tone reproduction continuously adjusts to the current view of the user. The overall performance is decreased only moderately, still allowing walkthroughs of large scenes.

The ability to perform interactive walkthroughs of global illumination solutions including glossy effects is a challenging open problem. In this paper we overcome certain limitations of previous approaches. We first introduce a novel, memory‐ and compute‐efficient representation of incoming illumination, in the context of a hierarchical radiance clustering algorithm. We then represent outgoing radiance with an adaptive hierarchical basis, in a manner suitable for interactive display. Using appropriate refinement and display strategies, we achieve walkthroughs of glossy solutions at interactive rates for non‐trivial scenes. In addition, our implementation has been developed to be portable and easily adaptable as an extension to existing, diffuse‐only, hierarchical radiosity systems. We present results of the implementation of glossy global illumination in two independent global illumination systems.

Clustering is a very efficient technique to apply finite element methods to the computation of radiosity solutions of complex scenes. Both computation time and memory consumption can be reduced dramatically by grouping the primitives of the input scene into a hierarchy of clusters and allowing for light exchange between all levels of this hierarchy. However, problems can arise due to clustering, when gross approximations about a cluster's content result in unsatisfactory solutions or unnecessary computations. In the clustering approach for diffuse global information described in the paper, light exchange between two objects-patches or clusters-is bounded by using geometrical and shading information provided by every object through a uniform interface. With this uniform view of various kinds of objects, comparable and reliable error bounds on the light exchange can be computed, which then guide a standard hierarchical radiosity algorithm.

Hierarchical radiosity with clustering has positioned itself as one of the most efficient algorithms for computing global illumination in non‐trivial environments. However, using hierarchical radiosity for complex scenes is still problematic due to the necessity of storing a large number of transport coefficients between surfaces in the form of links. In this paper, we eliminate the need for storage of links through the use of a modified shooting method for solving the radiosity equation. By distributing only unshot radiosity in each step of the iteration, the number of links decreases exponentially. Recomputing these links instead of storing them increases computation time, but reduces memory consumption dramatically. Caching may be used to reduce the time overhead. We analyze the error behavior of the new algorithm in comparison with the normal gathering approach for hierarchical radiosity. In particular, we consider the relation between the global error of a hierarchical radiosity solution and the local error threshold for each link.

Since the beginning of computer graphics, one of the primary goals has been to create convincingly realistic images of three-dimensional environments that would be impossible to distinguish from photographs of the real scene. The goal to create photo-realistic images has lead to the development of completely new software techniques for dealing with the inherent geometric and optical complexity of real world scenes. This paper gives an overview of advanced algorithms for photo-realistic rendering and in particular discusses hierarchical techniques for global illumination computations.

Computing global illumination by finite element techniques usually generates a piecewise constant approximation of the radiosity distribution on surfaces. Directly displaying such scenes generates artefacts due to discretization errors. We propose to remedy this drawback by considering the piecewise constant output to be samples of a (piecewise) smooth function in object space and reconstruct this function by applying a binary subdivision scheme. We design custom taylored subdivision schemes with quadratic precision for the efficient refinement of cell‐ or pixel‐type data. The technique naturally allows to reconstruct functions from non‐uniform samples which result from adaptive binary splitting of the original domain (quadtree). This type of output is produced, e.g., by hierarchical radiosity algorithms. The result of the subdivision process can be mapped as a texture on the respective surface patch which allows to exploit graphics hardware for considerably accelerating the display.

Traditionally, Radiosity algorithms have been restricted to scenes made from planar patches. Most algorithms for computing form factors and the subdivision criterion for hierarchical methods implicitly assume planar patches. In this paper, we present a new radiosity algorithm that is solely based on simple geometric information about surface elements, namely their bounding boxes and cone of normals. Using this information allows to compute efficient error bounds that can be used for the subdivision oracle and for computing the energy transfer. Due to the simple interface to geometric objects, our algorithm not only allows for computing illumination on general curved surfaces, but it can also be directly applied to a hieararchy of clusters. Several examples demonstrate the advantages of the new approach.