|
|
3D object classification using neural networks
Krabec, Miroslav ; Křivánek, Jaroslav (advisor) ; Šikudová, Elena (referee)
3D Object Classification Using Neural Networks Bc. Miroslav Krabec Classification of 3D objects is of great interest in the field of artificial intelligence. There are numerous approaches using artificial neural networks to address this problem. They differ mainly in the representation of the 3D model used as input and the network architecture. The goal of this thesis is to explore and test these approaches on publicly available datasets and subject them to independent comparison, which has not so far appeared in the literature. We provide a unified framework allowing to convert the data from common 3D formats. We train and test ten different network on the ModelNet40 and ShapeNetCore datasets. All the networks performed reasonably well in our tests, but we were generally unable to achieve the accuracies reported in the original papers. We suspect this could be due to extensive, albeit unreported, hyperparameter tuning by the authors of the original papers, suggesting this issue would benefit from further research. 1
|
|
|
Global exploration in Markov chain Monte Carlo methods for light transport simulation
Šik, Martin ; Křivánek, Jaroslav (advisor) ; Jakob, Wenzel (referee) ; Christensen, Per (referee)
Monte Carlo light transport simulation has become a de-facto standard tool for photorealistic rendering. However, the algorithms used by the current rendering systems are often ineffective, especially in scenes featuring light transport due to multiple highly glossy or specular interactions and complex visibility between the camera and light sources. It is therefore desirable to adopt more robust algorithms in practice. Light transport algorithms based on Markov chain Monte Carlo (MCMC) are known to be effective at sampling many different kinds of light transport paths even in the presence of complex visibility. However, the current MCMC algorithms often over-sample some of the paths while under-sampling or completely missing other paths. We attribute this behavior to insufficient global exploration of path space which leads to their unpredictable convergence and causes the occurrence of image artifacts. This in turn prohibits adoption of MCMC algorithms in practice. In this thesis we therefore focus on improving global exploration in MCMC algorithms for light transport simulation. First, we present a new MCMC algorithm that utilizes replica exchange to improve global exploration. To maximize efficiency of replica exchange we introduce tempering of the path space, which allows easier discovery of important...
|
|
|
Evaluation of Dynamic Range Reconstruction Approaches and a Mobile Application for HDR Photo Capture
Mirbauer, Martin ; Křivánek, Jaroslav (advisor) ; Šikudová, Elena (referee)
Digital photography became widespread with the global use of smartphones. However, most of the captured images do not fully use the camera capabilities by storing the captured photos in a format with limited dynamic range. The subject of dynamic range expansion and reconstruction has been researched since early 2000s and recently gave rise to several new reconstruction methods using convolutional neural networks (CNNs), whose performance has not yet been comprehensively compared. By implementing and using our dynamic range reconstruction evaluation framework we compare the reconstruction quality of individual CNN-based approaches. We also implement a mobile HDR camera application and evaluate the feasibility of running the best-performing reconstruction method directly on a mobile device.
|
|
|
Efficient rendering of fine structures on object surfaces
Brečka, Bohuš ; Křivánek, Jaroslav (advisor) ; Kondapaneni, Ivo (referee)
Current methods for realistic rendering approximate surface microstructure using a smooth normal distribution function. This approach is not sufficient for the rendering of shiny surfaces with details (such as scratches) visible under bright light in real world. It is possible to model surface structure with high-resolution normal maps, but this approach leads to unreasonable rendering times when used with modern rendering methods based on stochastic sampling. In this thesis, we explore some of the approaches specifically designed to address this problem. As a main topic we choose the algorithm proposed by Yen et al. [2016]. We analyse, implement it, compare it with other approaches and propose some improvements. As a part of this work we implement a rendering system based on the path tracing algorithm, which is used as an environment for testing and visualization of our results.
|
|
|
Procedural tree generation
Tázlar, Vojtěch ; Křivánek, Jaroslav (advisor) ; Kahoun, Martin (referee)
During 3D modelling of the real world, it is often necessary to place trees into the scene. However particular model may not suit users' needs and the use of a dedicated software can be inconvenient - due to complex control, impossible integration into the used system, model quality or price. At the time of writing this work, there exist a few professional high-end applications that are not suitable for use by a potential casual user, mainly because of their price. Then there are only freely available applications which are often obsolete and suffer from various shortcomings. Therefore, it makes sense to try to develop a low-end application, targeted for casual users, where the power of the tool is not as important as simplicity and nature of control and satisfactory results. We analyze ways of modelling trees using current applications and describe known algorithms and approaches to the generation of tree models. Based on this, we implement our own tool capable of generation of diverse tree types and adaptation of their models to obstacles in the scene. 1
|
|
|
Integration of the Corona renderer into the ArchiCAD software
Špaček, Jan ; Křivánek, Jaroslav (advisor) ; Vévoda, Petr (referee)
Integration of the Corona renderer into the ArchiCAD software Bachelor thesis - Jan Špaček, 2018 We present a prototype of a plug-in that integrates Corona Renderer into Archi- CAD. We strive to make Corona in ArchiCAD as easy to use as in other graphics applications, so the user interface is a major concern in our plug-in. Special focus was given to interactive rendering, which has proved to be a technical challenge in other host applications. We give a brief overview of the APIs that we use and describe the architecture and implementation of the plug-in. We also present the performance of interactive rendering on various scenes. 1
|
|
|
Image Denoising Using Weighted Local Regression
Šťasta, Jakub ; Křivánek, Jaroslav (advisor) ; Elek, Oskár (referee)
The problem of accurately simulating light transport using Monte Carlo integration can be very difficult. In particular, scenes with complex illumination effects or complex materials can cause a scene to converge very slowly and demand a lot of computational time. To overcome this problem, image denoising algorithms have become popular in recent years. In this work we first review known approaches to denoising and adaptive rendering. We implement one of the promising algorithm by Moon et al. [2014] in a commercial rendering system Corona Standalone Renderer, evaluate its performance, strengths and weaknesses on 14 test scenes. These include difficult to denoise and converge rendering effects such as fine sub-pixel geometry, participating media, extreme depth of field of highlights, motion blur, and others. We propose corrections which make the algorithm more stable and robust. We show that it is possible to denoise renderings with Linear Weighted Regression only using a CPU. However, still even after our propositions, it is not possible to filter scenes in a consistent manner without over-blurring or not filtering where desired.
|
|
|
Adjoint-Driven Importance Sampling in Light Transport Simulation
Vorba, Jiří ; Křivánek, Jaroslav (advisor) ; Keller, Alexander (referee) ; Wann Jensen, Henrik (referee)
Title: Adjoint-Driven Importance Sampling in Light Transport Simulation Author: RNDr. Jiří Vorba Department: Department of Software and Computer Science Education Supervisor: doc. Ing. Jaroslav Křivánek, Ph.D., Department of Software and Computer Science Education Abstract: Monte Carlo light transport simulation has recently been adopted by the movie industry as a standard tool for producing photo realistic imagery. As the industry pushes current technologies to the very edge of their possibilities, the unprecedented complexity of rendered scenes has underlined a fundamental weakness of MC light transport simulation: slow convergence in the presence of indirect illumination. The culprit of this poor behaviour is that the sam- pling schemes used in the state-of-the-art MC transport algorithms usually do not adapt to the conditions of rendered scenes. We base our work on the ob- servation that the vast amount of samples needed by these algorithms forms an abundant source of information that can be used to derive superior sampling strategies, tailored for a given scene. In the first part of this thesis, we adapt general machine learning techniques to train directional distributions for biasing scattering directions of camera paths towards incident illumination (radiance). Our approach allows progressive...
|
|
|
Procedural placement of 3D objects
Hojdar, Štěpán ; Křivánek, Jaroslav (advisor) ; Kahoun, Martin (referee)
3D modelling in computer graphics often requires placing a big number of objects into the scene. This may be tedious or even impossible if done manually. A few programs exist to perform this task automatically but most of them are either too slow to place the required number of objects or too difficult to use for a non-expert user. We expand the already existing Corona Scatter program which is fast and user friendly but lacks in terms of functionality. We implement scattering regular patterns, scattering along spline objects and the ability to use spline objects to locally modify the distribution properties.
|
|
|
Height map compression techniques
Lašan, Michal ; Kahoun, Martin (advisor) ; Křivánek, Jaroslav (referee)
The goal of this thesis is to design a suitable method for lossy compression of heightmap terrain data. This method should accept blocks of float samples of dimensions 2^n x 2^n as an input, for which it should be able to perform progressive decompression of mip-maps (lower-resolution representations). It should keep the reconstructed data within a certain maximum per-sample error bound for each mip-map level. This bound should be in the unit of meters and adjustable by the user. Given these constraints, it should be as efficient as possible. Our method is inspired by the second generation of progressive wavelet-based compression scheme modified to satisfy the~maximum-error constraint. We simplified this scheme by factoring out unnecessary computations in order to improve the efficiency. Our method can compress a 256x256 block in about 30 ms and decompress it in about 2 ms. Thanks to these attributes, the method can be used in a real-time planet renderer. It achieves the compression ratio of 37:1 on the whole Earth 90m/sample terrain dataset transformed and separated into square blocks, while respecting the maximum error of 5m. Powered by TCPDF (www.tcpdf.org)
|
guest ::
RSS feed.