[ { "id": "cardoso-thesis", "type_id": "phdthesis", "tu_id": null, "repositum_id": "20.500.12708/209309", "title": "Approaching Under-Explored Image-Space Problems with Optimization", "date": "2024-12-19", "abstract": "This doctoral dissertation delves into three distinct yet interconnected problems in the realm of interactive image-space computing in computer graphics, each of which has not been tackled by existing literature.The first problem centers on the prediction of visual error metrics in real-time applications, specifically in the context of content-adaptive shading and shading reuse. Utilizing convolutional neural networks, this research aims to estimate visual errors without requiring reference or rendered images. The models developed can account for 70%–90% of the variance and achieve computation times that are an order of magnitude faster than existing methods. This enables a balance between resource-saving and visual quality, particularly in deferred shading pipelines, and can achieve up to twice the performance compared to state-of-the-art methods depending on the portion of unseen image regions. The second problem focuses on the burgeoning field of light-field cameras and the challenges associated with depth prediction. This research argues for the refinement of cost volumes rather than depth maps to increase the accuracy of depth predictions. A set of cost-volume refinement algorithms is proposed, which dynamically operate at runtime to find optimal solutions, thereby enhancing the accuracy and reliability of depth estimation in light fields.The third problem tackles the labor-intensive nature of hand-drawn animation, specifically in the detailing of character eyes. An unsupervised network is introduced that blends inpainting and image-to-image translation techniques. This network employs a novel style-aware clustering method and a dual-discriminator optimization strategy with a triple-reconstruction loss. The result is an improvement in the level of detail and artistic consistency in hand-drawn animation, preferred over existing work 95.16% of the time according to a user study.Optimization techniques are the common thread that ties these problems together. While dynamic optimization at runtime is employed for cost volume refinement, deep-learning methods are used offline to train global solutions for the other two problems. This research not only fills gaps in the existing literature but also paves the way for future explorations in the field of computer graphics and optimization, offering new avenues for both academic research and practical applications.", "authors_et_al": false, "substitute": null, "main_image": null, "sync_repositum_override": "date", "repositum_presentation_id": null, "authors": [ 1639 ], "ac_number": "AC17414787", "date_end": "2024-12-19", "date_start": "2019-04", "doi": "10.34726/hss.2025.128664", "matrikelnr": "11937133", "open_access": "yes", "pages": "110", "reviewer_1": [ 1825 ], "reviewer_2": [ 5420 ], "rigorosum": "2024-12-19", "supervisor": [ 193 ], "research_areas": [ "Perception", "Rendering" ], "keywords": [ "variable-rate shading", "light-fields", "limited animation", "anime", "convolutional neural networks" ], "weblinks": [], "files": [ { "description": null, "filetitle": "thesis", "main_file": true, "use_in_gallery": false, "access": "public", "name": "cardoso-thesis-thesis.pdf", "type": "application/pdf", "size": 47447576, "path": "Publication:cardoso-thesis", "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-thesis/cardoso-thesis-thesis.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-thesis/cardoso-thesis-thesis:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-thesis/", "__class": "Publication" }, { "id": "cardoso-2024-r-c", "type_id": "inproceedings", "tu_id": null, "repositum_id": "20.500.12708/209907", "title": "Re:Draw - Context Aware Translation as a Controllable Method for Artistic Production", "date": "2024-08", "abstract": "We introduce context-aware translation, a novel method that combines the benefits of inpainting and image-to-image translation, respecting simultaneously the original input and contextual relevance – where existing methods fall short. By doing so, our method opens new avenues for the controllable use of AI within artistic creation, from animation to digital art.\nAs an use case, we apply our method to redraw any hand-drawn animated character eyes based on any design specifications – eyes serve as a focal point that captures viewer attention and conveys a range of emotions; however, the labor-intensive na-\nture of traditional animation often leads to compromises in the complexity and consistency of eye design. Furthermore, we remove the need for production data for training and introduce a new character recognition method that surpasses existing work\nby not requiring fine-tuning to specific productions.\nThis proposed use case could help maintain consistency throughout production and unlock bolder and\nmore detailed design choices without the production cost drawbacks. A user study shows contextaware translation is preferred over existing work 95.16% of the time.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "image", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 343, "image_height": 294, "name": "cardoso-2024-r-c-image.bmp", "type": "image/bmp", "size": 303462, "path": "Publication:cardoso-2024-r-c", "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-image.bmp", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-image:thumb{{size}}.png" }, "sync_repositum_override": "date,projects", "repositum_presentation_id": null, "authors": [ 1639, 5437, 1519, 193 ], "booktitle": "Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-24)", "date_from": "2024-08-03", "date_to": "2024-08-09", "doi": "10.24963/ijcai.2024/842", "event": "33rd International Joint Conference on Artificial Intelligence (IJCAI 2024)", "isbn": "978-1-956792-04-1", "lecturer": [ 1639 ], "location": "Jeju Island", "pages": "9", "pages_from": "7609", "pages_to": "7617", "publisher": "International Joint Conferences on Artificial Intelligence", "research_areas": [ "Rendering" ], "keywords": [ "Application domains: Images, movies and visual arts", "Application domains: Computer Graphics and Animation", "Methods and resources: AI systems for collaboration and co-creation", "Methods and resources: Machine learning, deep learning, neural models, reinforcement learning", "Theory and philosophy of arts and creativity in AI systems: Social (multi-agent) creativity and human-computer co-creation" ], "weblinks": [], "files": [ { "description": null, "filetitle": "image", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 343, "image_height": 294, "name": "cardoso-2024-r-c-image.bmp", "type": "image/bmp", "size": 303462, "path": "Publication:cardoso-2024-r-c", "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-image.bmp", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-image:thumb{{size}}.png" }, { "description": null, "filetitle": "paper", "main_file": true, "use_in_gallery": false, "access": "public", "name": "cardoso-2024-r-c-paper.pdf", "type": "application/pdf", "size": 2952059, "path": "Publication:cardoso-2024-r-c", "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/cardoso-2024-r-c-paper:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2024/cardoso-2024-r-c/", "__class": "Publication" }, { "id": "cardoso-2022-rtpercept", "type_id": "journalpaper", "tu_id": null, "repositum_id": "20.500.12708/142206", "title": "Training and Predicting Visual Error for Real-Time Applications", "date": "2022-05", "abstract": "Visual error metrics play a fundamental role in the quantification of perceived image similarity. Most recently, use cases for them in real-time applications have emerged, such as content-adaptive shading and shading reuse to increase performance and improve efficiency. A wide range of different metrics has been established, with the most sophisticated being capable of capturing the perceptual characteristics of the human visual system. However, their complexity, computational expense, and reliance on reference images to compare against prevent their generalized use in real-time, restricting such applications to using only the simplest available metrics.\n\nIn this work, we explore the abilities of convolutional neural networks to predict a variety of visual metrics without requiring either reference or rendered images. Specifically, we train and deploy a neural network to estimate the visual error resulting from reusing shading or using reduced shading rates. The resulting models account for 70%--90% of the variance while achieving up to an order of magnitude faster computation times. Our solution combines image-space information that is readily available in most state-of-the-art deferred shading pipelines with reprojection from previous frames to enable an adequate estimate of visual errors, even in previously unseen regions. We describe a suitable convolutional network architecture and considerations for data preparation for training. We demonstrate the capability of our network to predict complex error metrics at interactive rates in a real-time application that implements content-adaptive shading in a deferred pipeline. Depending on the portion of unseen image regions, our approach can achieve up to 2x performance compared to state-of-the-art methods.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 1920, "image_height": 1088, "name": "cardoso-2022-rtpercept-teaser.png", "type": "image/png", "size": 2493649, "path": "Publication:cardoso-2022-rtpercept", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1639, 1650, 868, 1921, 193 ], "cfp": { "name": "cfp.pdf", "type": "application/pdf", "error": "0", "size": "1227280", "orig_name": "cfp.pdf", "ext": "pdf" }, "date_from": "2022-05-03", "date_to": "2022-05-05", "doi": "10.1145/3522625", "event": "ACM SIGGRAPH Symposium on Interactive 3D Graphics and Games", "issn": "2577-6193", "journal": "Proceedings of the ACM on Computer Graphics and Interactive Techniques", "lecturer": [ 1639 ], "location": "online", "number": "1", "open_access": "yes", "pages": "17", "pages_from": "1", "pages_to": "17", "publisher": "Association for Computing Machinery", "volume": "5", "research_areas": [ "Perception", "Rendering" ], "keywords": [ "perceptual error", "variable rate shading", "real-time" ], "weblinks": [ { "href": "https://jaliborc.github.io/rt-percept/", "caption": "Paper Website", "description": null, "main_file": 1 } ], "files": [ { "description": null, "filetitle": "paper", "main_file": true, "use_in_gallery": false, "access": "public", "name": "cardoso-2022-rtpercept-paper.pdf", "type": "application/pdf", "size": 54709850, "path": "Publication:cardoso-2022-rtpercept", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-paper:thumb{{size}}.png" }, { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 1920, "image_height": 1088, "name": "cardoso-2022-rtpercept-teaser.png", "type": "image/png", "size": 2493649, "path": "Publication:cardoso-2022-rtpercept", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/cardoso-2022-rtpercept-teaser:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2022/cardoso-2022-rtpercept/", "__class": "Publication" }, { "id": "celarek-2022-gmcn", "type_id": "inproceedings", "tu_id": null, "repositum_id": "20.500.12708/188182", "title": "Gaussian Mixture Convolution Networks", "date": "2022-04", "abstract": "This paper proposes a novel method for deep learning based on the analytical convolution of multidimensional Gaussian mixtures.\nIn contrast to tensors, these do not suffer from the curse of dimensionality and allow for a compact representation, as data is only stored where details exist.\nConvolution kernels and data are Gaussian mixtures with unconstrained weights, positions, and covariance matrices.\nSimilar to discrete convolutional networks, each convolution step produces several feature channels, represented by independent Gaussian mixtures.\nSince traditional transfer functions like ReLUs do not produce Gaussian mixtures, we propose using a fitting of these functions instead.\nThis fitting step also acts as a pooling layer if the number of Gaussian components is reduced appropriately.\nWe demonstrate that networks based on this architecture reach competitive accuracy on Gaussian mixtures fitted to the MNIST and ModelNet data sets.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 1500, "image_height": 1367, "name": "celarek-2022-gmcn-teaser.png", "type": "image/png", "size": 720960, "path": "Publication:celarek-2022-gmcn", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1013, 1919, 1650, 951, 193 ], "booktitle": "The Tenth International Conference on Learning Representations (ICLR 2022)", "cfp": { "name": "ICLR2022_Call for Papers.pdf", "type": "application/pdf", "error": "0", "size": "32122", "orig_name": "ICLR2022_Call for Papers.pdf", "ext": "pdf" }, "event": "ICLR | 2022", "lecturer": [ 1013 ], "open_access": "yes", "pages_from": "1", "pages_to": "23", "publisher": "OpenReview.org", "research_areas": [ "Geometry" ], "keywords": [], "weblinks": [ { "href": "https://github.com/cg-tuwien/Gaussian-Mixture-Convolution-Networks", "caption": "Code on github", "description": null, "main_file": 1 }, { "href": "https://openreview.net/forum?id=Oxeka7Z7Hor", "caption": "Paper on OpenReview", "description": null, "main_file": 0 } ], "files": [ { "description": null, "filetitle": "paper", "main_file": true, "use_in_gallery": false, "access": "public", "name": "celarek-2022-gmcn-paper.pdf", "type": "application/pdf", "size": 5943864, "path": "Publication:celarek-2022-gmcn", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-paper:thumb{{size}}.png" }, { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 1500, "image_height": 1367, "name": "celarek-2022-gmcn-teaser.png", "type": "image/png", "size": 720960, "path": "Publication:celarek-2022-gmcn", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/celarek-2022-gmcn-teaser:thumb{{size}}.png" } ], "projects_workgroups": [ "rend", "EVOCATION", "3DSpatialization" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2022/celarek-2022-gmcn/", "__class": "Publication" }, { "id": "FRAISS-2022-CGMM", "type_id": "masterthesis", "tu_id": null, "repositum_id": "20.500.12708/19748", "title": "Construction and Visualization of Gaussian Mixture Models from Point Clouds for 3D Object Representation", "date": "2022-03-10", "abstract": "Point clouds are a common representation of three-dimensional shapes in computer graphics\nand 3D-data processing. However, in some applications, other representations are more useful.\nGaussian Mixture Models (GMMs) can be used as such an alternative representation. A GMM\nis a convex sum of normal distributions, which aims to describe a point cloud’s density. In\nthis thesis, we investigate both visualization and construction of GMMs. For visualization,\nwe have implemented a tool that enables both isoellipsoid and density visualization of GMMs.\nWe describe the mathematical backgrounds, the algorithms, and our implementation of this\ntool. Regarding GMM construction, we investigate several algorithms used in previous papers\nfor constructing GMMs for 3D-data processing tasks. We present our implementations of the\nexpectation-maximization (EM) algorithm and top-down HEM. Additionally, we have adapted\nthe implementation of geometrically regularized bottom-up HEM to produce a fixed number of\nGaussians. We evaluate these three algorithms in terms of the quality of their generated GMMs.\nIn many cases, the statistical likelihood, which is maximized by the EM algorithm, is not a\nreliable indicator for a GMM’s quality. Therefore, we instead rely on the reconstruction error of a\nreconstructed point cloud based on the Chamfer distance. Additionally, we provide metrics for\nmeasuring the reconstructed point cloud’s uniformity and the GMM’s variation of Gaussians. We\ndemonstrate that EM provides the best results in terms of these metrics. Top-down HEM is a fast\nalternative, and can produce even better results when using fewer input points. The results of\ngeometrically regularized bottom-up HEM are inferior for lower numbers of Gaussians but it can\ncreate good GMMs consisting of high numbers of Gaussians very eciently.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "image", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 1380, "image_height": 1850, "name": "FRAISS-2022-CGMM-image.jpg", "type": "image/jpeg", "size": 376815, "path": "Publication:FRAISS-2022-CGMM", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-image.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-image:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1495 ], "co_supervisor": [ 1013 ], "date_end": "2022-03-10", "date_start": "2021-01-01", "diploma_examina": "2022-03-10", "matrikelnr": "01425602", "supervisor": [ 193 ], "research_areas": [ "Rendering" ], "keywords": [], "weblinks": [], "files": [ { "description": null, "filetitle": "image", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 1380, "image_height": 1850, "name": "FRAISS-2022-CGMM-image.jpg", "type": "image/jpeg", "size": 376815, "path": "Publication:FRAISS-2022-CGMM", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-image.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-image:thumb{{size}}.png" }, { "description": null, "filetitle": "Master thesis", "main_file": true, "use_in_gallery": false, "access": "public", "name": "FRAISS-2022-CGMM-Master thesis.pdf", "type": "application/pdf", "size": 20842982, "path": "Publication:FRAISS-2022-CGMM", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-Master thesis.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-Master thesis:thumb{{size}}.png" }, { "description": null, "filetitle": "poster", "main_file": true, "use_in_gallery": false, "access": "public", "name": "FRAISS-2022-CGMM-poster.pdf", "type": "application/pdf", "size": 4483975, "path": "Publication:FRAISS-2022-CGMM", "url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-poster.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/FRAISS-2022-CGMM-poster:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2022/FRAISS-2022-CGMM/", "__class": "Publication" }, { "id": "cardoso-2021-cost", "type_id": "inproceedings", "tu_id": 301688, "repositum_id": null, "title": "Cost Volume Refinement for Depth Prediction", "date": "2021-01-10", "abstract": "Light-field cameras are becoming more popular in\nthe consumer market. Their data redundancy allows, in theory,\nto accurately refocus images after acquisition and to predict the\ndepth of each point visible from the camera. Combined, these\ntwo features allow for the generation of full-focus images, which\nis impossible in traditional cameras.\nMultiple methods for depth prediction from light fields (or\nstereo) have been proposed over the years. A large subset of\nthese methods relies on cost-volume estimates – 3D objects where\neach layer represents a heuristic of whether each point in the\nimage is at a certain distance from the camera. Generally, this\nvolume is used to regress a depth map, which is then refined\nfor better results. In this paper, we argue that refining the cost\nvolumes is superior to refining the depth maps in order to further\nincrease the accuracy of depth predictions. We propose a set of\ncost-volume refinement algorithms and show their effectiveness.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "header", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "cardoso-2021-cost-header.jpg", "type": "image/jpeg", "size": 34129, "path": "Publication:cardoso-2021-cost", "url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-header.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-header:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1639, 1801, 193 ], "booktitle": "Proceedings of the 25th International Conference on Pattern Recognition", "cfp": { "name": "1.html", "type": "text/html", "error": "0", "size": "51497", "orig_name": "1.html", "ext": "html" }, "date_from": "2021-01-10", "date_to": "2021-01-15", "doi": "10.1109/ICPR48806.2021.9412730", "event": "25th International Conference on Pattern Recognition (ICPR)", "isbn": "978-1-7281-8809-6", "lecturer": [ 1639 ], "location": "Milan, Italy", "open_access": "yes", "pages_from": "354", "pages_to": "361", "publisher": "IEEE", "research_areas": [ "Rendering" ], "keywords": [ "depth reconstruction", "light fields", "cost volumes" ], "weblinks": [], "files": [ { "description": "Acknowledgements section added.", "filetitle": "amended-paper", "main_file": true, "use_in_gallery": false, "access": "public", "name": "cardoso-2021-cost-amended-paper.pdf", "type": "application/pdf", "size": 11789393, "path": "Publication:cardoso-2021-cost", "url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-amended-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-amended-paper:thumb{{size}}.png" }, { "description": null, "filetitle": "header", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "cardoso-2021-cost-header.jpg", "type": "image/jpeg", "size": 34129, "path": "Publication:cardoso-2021-cost", "url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-header.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-header:thumb{{size}}.png" }, { "description": null, "filetitle": "original-paper", "main_file": false, "use_in_gallery": false, "access": "public", "name": "cardoso-2021-cost-original-paper.pdf", "type": "application/pdf", "size": 11772629, "path": "Publication:cardoso-2021-cost", "url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-original-paper.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/cardoso-2021-cost-original-paper:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2021/cardoso-2021-cost/", "__class": "Publication" }, { "id": "hanko-2019-ani", "type_id": "bachelorthesis", "tu_id": null, "repositum_id": null, "title": "Higher Hand-Drawn Detail Quality using Convolutional Assistant", "date": "2020-04", "abstract": "The field of research in the use of neural networks to help artists or advance 2D animation\nis very underdeveloped. Most of the little research that is done does not even ask questions\nthat are relevant for animators but is done in a pure research mindset. We, however,\ntried to find a problem that would actually be relevant in the animation industry and\ncame up with the idea of enhancing the feature quality of poorly drawn features in 2D\nanimation. The basis for this idea is that, as a cost and time-saving measure, in 2d\nanimation features are often drawn in different levels of detail depending on the current\nfocus of the scene and other factors. The focus will thereby lie on the enhancement of\ncharacters’ eyes with the idea that other features could be done in a similar way in future\nwork. To achieve this quality enhancing we train the FUNIT network on a\nmanually created dataset consisting of crops of eyes from different characters in different\nquality with the goal that it will be able to consistently transform low-quality eye images\ninto high-quality eye images", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "hanko-2019-ani-teaser.jpg", "type": "image/jpeg", "size": 45990, "path": "Publication:hanko-2019-ani", "url": "https://www.cg.tuwien.ac.at/research/publications/2020/hanko-2019-ani/hanko-2019-ani-teaser.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2020/hanko-2019-ani/hanko-2019-ani-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1744 ], "date_end": "2020-04", "date_start": "2019-10", "matrikelnr": "01625726", "supervisor": [ 1639, 193 ], "research_areas": [], "keywords": [], "weblinks": [], "files": { "1": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "hanko-2019-ani-teaser.jpg", "type": "image/jpeg", "size": 45990, "path": "Publication:hanko-2019-ani", "url": "https://www.cg.tuwien.ac.at/research/publications/2020/hanko-2019-ani/hanko-2019-ani-teaser.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2020/hanko-2019-ani/hanko-2019-ani-teaser:thumb{{size}}.png" } }, "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2020/hanko-2019-ani/", "__class": "Publication" }, { "id": "wieser-2019-ani", "type_id": "bachelorthesis", "tu_id": null, "repositum_id": null, "title": "Classification of Production Ready 2D Animation using Contour and Distance Fields", "date": "2019-12", "abstract": "Image classification is one of the most common use cases of Convolutional Neural Networks. In this thesis, our goal is to increase the accuracy of a neural network classifier for frames of production ready 2D animations and to create a model from a dataset with high accuracy for classification. This can be seen as groundwork for future work that applies neural networks on production ready 2D animation data, by reusing and tweaking the model for different applications.\n\nWe compare training a neural network with the color channels of images to training with\ngrayscale images, predicted contours or distance fields generated from those contours.\nFurthermore, different combinations of the data will be used to evaluate the best option.\nThis means that the comparison of the accuracy not only includes color data compared\nto color with contours and distance fields but every combination of the aforementioned\nfour types of input.", "authors_et_al": false, "substitute": null, "main_image": { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "wieser-2019-ani-teaser.jpg", "type": "image/jpeg", "size": 14419, "path": "Publication:wieser-2019-ani", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/wieser-2019-ani/wieser-2019-ani-teaser.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/wieser-2019-ani/wieser-2019-ani-teaser:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1657 ], "date_end": "2019-12", "date_start": "2019-06", "matrikelnr": "01633066", "supervisor": [ 193, 1639 ], "research_areas": [], "keywords": [], "weblinks": [], "files": [ { "description": null, "filetitle": "teaser", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 217, "image_height": 145, "name": "wieser-2019-ani-teaser.jpg", "type": "image/jpeg", "size": 14419, "path": "Publication:wieser-2019-ani", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/wieser-2019-ani/wieser-2019-ani-teaser.jpg", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/wieser-2019-ani/wieser-2019-ani-teaser:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2019/wieser-2019-ani/", "__class": "Publication" }, { "id": "celarek_adam-2019-qelta", "type_id": "journalpaper", "tu_id": 282852, "repositum_id": null, "title": "Quantifying the Error of Light Transport Algorithms", "date": "2019-07", "abstract": "This paper proposes a new methodology for measuring the error of unbiased physically based rendering algorithms. The current state of the art includes mean squared error (MSE) based metrics and visual comparisons of equal-time renderings of competing algorithms. Neither is satisfying as MSE does not describe behavior and can exhibit significant variance, and visual comparisons are inherently subjective. Our contribution is two-fold: First, we propose to compute many short renderings instead of a single long run and use the short renderings to estimate MSE expectation and variance as well as per-pixel standard deviation. An algorithm that achieves good results in most runs, but with occasional outliers is essentially unreliable, which we wish to quantify numerically. We use per-pixel standard deviation to identify problematic lighting effects of rendering algorithms. The second contribution is the error spectrum ensemble (ESE), a tool for measuring the distribution of error over frequencies. The ESE serves two purposes: It reveals correlation between pixels and can be used to detect outliers, which offset the amount of error substantially.", "authors_et_al": false, "substitute": null, "main_image": { "description": "Error spectrum ensemble", "filetitle": "image", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 594, "image_height": 372, "name": "celarek_adam-2019-qelta-image.png", "type": "image/png", "size": 39262, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-image.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-image:thumb{{size}}.png" }, "sync_repositum_override": null, "repositum_presentation_id": null, "authors": [ 1013, 1667, 193, 1666 ], "cfp": { "name": "EGSR-2019_Call-for-Papers.pdf", "type": "application/pdf", "error": "0", "size": "174109", "orig_name": "EGSR-2019_Call-for-Papers.pdf", "ext": "pdf" }, "date_from": "2019-07-10", "date_to": "2019-07-12", "doi": "10.1111/cgf.13775", "event": "Eurographics Symposium on Rendering 2019", "journal": "Computer Graphics Forum", "lecturer": [ 1013 ], "number": "4", "open_access": "yes", "pages_from": "111", "pages_to": "121", "publisher": "The Eurographics Association and John Wiley & Sons Ltd.", "volume": "38", "research_areas": [ "Rendering" ], "keywords": [ "measuring error", "light transport", "global illumination" ], "weblinks": [ { "href": "https://github.com/cg-tuwien/Quantifying-the-Error-of-Light-Transport-Algorithms", "caption": "Git repository", "description": "Git repository with the implementation of the error spectrum ensemble and short rendering generator.", "main_file": 1 } ], "files": [ { "description": "Error spectrum ensemble", "filetitle": "image", "main_file": false, "use_in_gallery": false, "access": "public", "image_width": 594, "image_height": 372, "name": "celarek_adam-2019-qelta-image.png", "type": "image/png", "size": 39262, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-image.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-image:thumb{{size}}.png" }, { "description": "Paper", "filetitle": "paper_preprint", "main_file": true, "use_in_gallery": false, "access": "public", "name": "celarek_adam-2019-qelta-paper_preprint.pdf", "type": "application/pdf", "size": 2849560, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-paper_preprint.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-paper_preprint:thumb{{size}}.png" }, { "description": "Slides from the presentation at EGSR", "filetitle": "presentation", "main_file": false, "use_in_gallery": false, "access": "public", "name": "celarek_adam-2019-qelta-presentation.pdf", "type": "application/pdf", "size": 3561061, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-presentation.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-presentation:thumb{{size}}.png" }, { "description": "Additional examples, experiments and similar.", "filetitle": "supplemental_material", "main_file": false, "use_in_gallery": false, "access": "public", "name": "celarek_adam-2019-qelta-supplemental_material.pdf", "type": "application/pdf", "size": 81913248, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-supplemental_material.pdf", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-supplemental_material:thumb{{size}}.png" }, { "description": "Teaser", "filetitle": "teaser", "main_file": false, "use_in_gallery": true, "access": "public", "image_width": 1826, "image_height": 484, "name": "celarek_adam-2019-qelta-teaser.png", "type": "image/png", "size": 407655, "path": "Publication:celarek_adam-2019-qelta", "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-teaser.png", "thumb_image_sizes": [ 16, 64, 100, 175, 300, 600 ], "thumb_url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/celarek_adam-2019-qelta-teaser:thumb{{size}}.png" } ], "projects_workgroups": [ "EVOCATION", "OpenData" ], "url": "https://www.cg.tuwien.ac.at/research/publications/2019/celarek_adam-2019-qelta/", "__class": "Publication" } ]