O infravermelho termal (TIR - Thermal InfraRed) é uma porção do espectro eletromagnético com várias aplicações no Sensoriamento Remoto (SR), tais como: geologia, climatologia, análises de processos biológicos, análises geofísicas, avaliação de desastres e detecção de mudanças, entre outras. No TIR a emissão de radiação dos alvos é uma função não linear de duas variáveis, a emissividade e a temperatura do alvo, e a principal dificuldade é calcular/estimar tais variáveis separadamente e de forma confiável. Vários métodos foram desenvolvidos nas últimas décadas para mitigar esta indeterminação, mas independente do método todos tem a mesma deficiência, são desenvolvidos para aplicações específicas como o tipo de sensor, tipo de estudo, o alvo em análise, o número de alvos, tipo de clima, entre outros. Desta forma, o método a ser aplicado depende do estudo em questão, e para obter melhores resultados deve-se escolher o método que melhor se aplica ao problema estudado pelo analista. Neste trabalho, se propõe uma abordagem alternativa para estimar a temperatura e, portanto, a emissividade de um alvo em particular. A abordagem consiste em gerar regressões, em determinados comprimentos de onda, a partir da função linearizada da radiância para dados de laboratório de uma amostra de quartzo em diferentes temperaturas, medida sob condições controladas de humidade e temperatura do ambiente. As regressões visam modelar a variação na temperatura devido as variações na radiância do alvo, de modo a estimar a temperatura a partir da radiância em determinado comprimento de onda e sem o conhecimento prévio da emissividade do alvo. Os dados de laboratório foram divididos em dois grupos, treinamento e controle, no grupo de treinamento várias regressões polinomiais foram aplicados enquanto os dados de controle serviram para validar e avaliar as regressões. Foram realizados 5 experimentos: 1) dados de laboratório em comprimentos de onda específicos, 2) nos comprimentos de ondas centrais das bandas TIR-ASTER, 3) nas simulações das bandas TIR-ASTER, 4) com a simulação da atmosfera (seca e úmida) para as bandas simuladas TIRASTER e 5) numa imagem L1B TIR-ASTER da região de estudo e validado com o produto AST08. Como resultado, foi possível estimar a temperatura com erros menores que 0.2K para os dados de laboratório e com erro médio menor que 1.5K para imagens LIB TIR-ASTER. Além disso, o método requer apenas uma banda espectral na imagem, viabilizando sua aplicação em sensores termais monoespectrais. Resultados satisfatórios foram obtidos com uma regressão linear simples, e melhoram ao aumentar o comprimento de onda. No entanto, aumentando o comprimento de onda e, simultaneamente, o grau do polinômio da regressão os resultados também melhoram com relação a regressão linear, porém não são significativos, e desta forma o ajuste linear é a melhor opção. Desta forma, o método proposto se mostrou promissor, sinalizando que futuras pesquisas são necessárias. / The thermal infrared (TIR) is a portion of the electromagnetic spectrum with multiple remote Sensing applications in the field of geology, climatology, biological processes analysis, geophysical analysis, disaster assessment, change detection and many others. In TIR, radiation emission of the target is a nonlinear function of two unknowns – the emissivity and the temperature, and the main difficulty is to calculate/estimate these two variables separately and reliably. Several methods have been developed in the recent decades to mitigate this problem. However, regardless of the method, all have developed similar incapacities for specific applications such as the type of sensor, study type, the target in question, the number of targets, type of weather, among others. Thus, the method to be applied depends on the study in question and the best results can be reached choosing the best fit method for that problem. In this work, we propose an alternative approach for estimating the temperature, and therefore the emissivity, of a particular target. The approach consists of generating statistical regressions in some wavelengths from linearized radiance function of laboratory data from a quartz sample at different temperatures, measured under controlled conditions of humidity and room temperature. The aim of regressions is to model the variation in temperature due to the variations in the radiance of the target in order to estimate the temperature from radiance data on a certain wavelength and without prior knowledge of the target emissivity. Laboratory datasets were divided into two groups - training and control. In the training group, several polynomial regressions were applied while the control group served to validate and evaluate the regressions. Five experiments were performed: (1) laboratory data at specific wavelengths (2) the central wave lengths of ASTER-TIR bands (3) simulations of ASTER-TIR bands (4) simulation of the atmosphere (dry and wet) for simulated bands of ASTER-TIR and (5) an image L1B ASTER-TIR of the study area validated with the AST08 product. As a result, it was possible to estimate the temperature with errors less than 0.2K from laboratory data and with mean error less than 1.5K from L1B ASTER-TIR images. Furthermore, the method requires only a spectral band in the image, enabling their application in monospectral thermal sensors. Satisfactory results were obtained with a simple linear regression and improved by increasing the wavelength. However, increasing the wavelength and, simultaneously, the degree of polynomial regression the results also improve with respect to linear regression results, but this improvement is insignificant, and thus the linear fit is the best option. Thus, the proposed method has shown promise, signaling that further research is needed.

O infravermelho termal (TIR - Thermal InfraRed) é uma porção do espectro eletromagnético com várias aplicações no Sensoriamento Remoto (SR), tais como: geologia, climatologia, análises de processos biológicos, análises geofísicas, avaliação de desastres e detecção de mudanças, entre outras. No TIR a emissão de radiação dos alvos é uma função não linear de duas variáveis, a emissividade e a temperatura do alvo, e a principal dificuldade é calcular/estimar tais variáveis separadamente e de forma confiável. Vários métodos foram desenvolvidos nas últimas décadas para mitigar esta indeterminação, mas independente do método todos tem a mesma deficiência, são desenvolvidos para aplicações específicas como o tipo de sensor, tipo de estudo, o alvo em análise, o número de alvos, tipo de clima, entre outros. Desta forma, o método a ser aplicado depende do estudo em questão, e para obter melhores resultados deve-se escolher o método que melhor se aplica ao problema estudado pelo analista. Neste trabalho, se propõe uma abordagem alternativa para estimar a temperatura e, portanto, a emissividade de um alvo em particular. A abordagem consiste em gerar regressões, em determinados comprimentos de onda, a partir da função linearizada da radiância para dados de laboratório de uma amostra de quartzo em diferentes temperaturas, medida sob condições controladas de humidade e temperatura do ambiente. As regressões visam modelar a variação na temperatura devido as variações na radiância do alvo, de modo a estimar a temperatura a partir da radiância em determinado comprimento de onda e sem o conhecimento prévio da emissividade do alvo. Os dados de laboratório foram divididos em dois grupos, treinamento e controle, no grupo de treinamento várias regressões polinomiais foram aplicados enquanto os dados de controle serviram para validar e avaliar as regressões. Foram realizados 5 experimentos: 1) dados de laboratório em comprimentos de onda específicos, 2) nos comprimentos de ondas centrais das bandas TIR-ASTER, 3) nas simulações das bandas TIR-ASTER, 4) com a simulação da atmosfera (seca e úmida) para as bandas simuladas TIRASTER e 5) numa imagem L1B TIR-ASTER da região de estudo e validado com o produto AST08. Como resultado, foi possível estimar a temperatura com erros menores que 0.2K para os dados de laboratório e com erro médio menor que 1.5K para imagens LIB TIR-ASTER. Além disso, o método requer apenas uma banda espectral na imagem, viabilizando sua aplicação em sensores termais monoespectrais. Resultados satisfatórios foram obtidos com uma regressão linear simples, e melhoram ao aumentar o comprimento de onda. No entanto, aumentando o comprimento de onda e, simultaneamente, o grau do polinômio da regressão os resultados também melhoram com relação a regressão linear, porém não são significativos, e desta forma o ajuste linear é a melhor opção. Desta forma, o método proposto se mostrou promissor, sinalizando que futuras pesquisas são necessárias. / The thermal infrared (TIR) is a portion of the electromagnetic spectrum with multiple remote Sensing applications in the field of geology, climatology, biological processes analysis, geophysical analysis, disaster assessment, change detection and many others. In TIR, radiation emission of the target is a nonlinear function of two unknowns – the emissivity and the temperature, and the main difficulty is to calculate/estimate these two variables separately and reliably. Several methods have been developed in the recent decades to mitigate this problem. However, regardless of the method, all have developed similar incapacities for specific applications such as the type of sensor, study type, the target in question, the number of targets, type of weather, among others. Thus, the method to be applied depends on the study in question and the best results can be reached choosing the best fit method for that problem. In this work, we propose an alternative approach for estimating the temperature, and therefore the emissivity, of a particular target. The approach consists of generating statistical regressions in some wavelengths from linearized radiance function of laboratory data from a quartz sample at different temperatures, measured under controlled conditions of humidity and room temperature. The aim of regressions is to model the variation in temperature due to the variations in the radiance of the target in order to estimate the temperature from radiance data on a certain wavelength and without prior knowledge of the target emissivity. Laboratory datasets were divided into two groups - training and control. In the training group, several polynomial regressions were applied while the control group served to validate and evaluate the regressions. Five experiments were performed: (1) laboratory data at specific wavelengths (2) the central wave lengths of ASTER-TIR bands (3) simulations of ASTER-TIR bands (4) simulation of the atmosphere (dry and wet) for simulated bands of ASTER-TIR and (5) an image L1B ASTER-TIR of the study area validated with the AST08 product. As a result, it was possible to estimate the temperature with errors less than 0.2K from laboratory data and with mean error less than 1.5K from L1B ASTER-TIR images. Furthermore, the method requires only a spectral band in the image, enabling their application in monospectral thermal sensors. Satisfactory results were obtained with a simple linear regression and improved by increasing the wavelength. However, increasing the wavelength and, simultaneously, the degree of polynomial regression the results also improve with respect to linear regression results, but this improvement is insignificant, and thus the linear fit is the best option. Thus, the proposed method has shown promise, signaling that further research is needed.

O infravermelho termal (TIR - Thermal InfraRed) é uma porção do espectro eletromagnético com várias aplicações no Sensoriamento Remoto (SR), tais como: geologia, climatologia, análises de processos biológicos, análises geofísicas, avaliação de desastres e detecção de mudanças, entre outras. No TIR a emissão de radiação dos alvos é uma função não linear de duas variáveis, a emissividade e a temperatura do alvo, e a principal dificuldade é calcular/estimar tais variáveis separadamente e de forma confiável. Vários métodos foram desenvolvidos nas últimas décadas para mitigar esta indeterminação, mas independente do método todos tem a mesma deficiência, são desenvolvidos para aplicações específicas como o tipo de sensor, tipo de estudo, o alvo em análise, o número de alvos, tipo de clima, entre outros. Desta forma, o método a ser aplicado depende do estudo em questão, e para obter melhores resultados deve-se escolher o método que melhor se aplica ao problema estudado pelo analista. Neste trabalho, se propõe uma abordagem alternativa para estimar a temperatura e, portanto, a emissividade de um alvo em particular. A abordagem consiste em gerar regressões, em determinados comprimentos de onda, a partir da função linearizada da radiância para dados de laboratório de uma amostra de quartzo em diferentes temperaturas, medida sob condições controladas de humidade e temperatura do ambiente. As regressões visam modelar a variação na temperatura devido as variações na radiância do alvo, de modo a estimar a temperatura a partir da radiância em determinado comprimento de onda e sem o conhecimento prévio da emissividade do alvo. Os dados de laboratório foram divididos em dois grupos, treinamento e controle, no grupo de treinamento várias regressões polinomiais foram aplicados enquanto os dados de controle serviram para validar e avaliar as regressões. Foram realizados 5 experimentos: 1) dados de laboratório em comprimentos de onda específicos, 2) nos comprimentos de ondas centrais das bandas TIR-ASTER, 3) nas simulações das bandas TIR-ASTER, 4) com a simulação da atmosfera (seca e úmida) para as bandas simuladas TIRASTER e 5) numa imagem L1B TIR-ASTER da região de estudo e validado com o produto AST08. Como resultado, foi possível estimar a temperatura com erros menores que 0.2K para os dados de laboratório e com erro médio menor que 1.5K para imagens LIB TIR-ASTER. Além disso, o método requer apenas uma banda espectral na imagem, viabilizando sua aplicação em sensores termais monoespectrais. Resultados satisfatórios foram obtidos com uma regressão linear simples, e melhoram ao aumentar o comprimento de onda. No entanto, aumentando o comprimento de onda e, simultaneamente, o grau do polinômio da regressão os resultados também melhoram com relação a regressão linear, porém não são significativos, e desta forma o ajuste linear é a melhor opção. Desta forma, o método proposto se mostrou promissor, sinalizando que futuras pesquisas são necessárias. / The thermal infrared (TIR) is a portion of the electromagnetic spectrum with multiple remote Sensing applications in the field of geology, climatology, biological processes analysis, geophysical analysis, disaster assessment, change detection and many others. In TIR, radiation emission of the target is a nonlinear function of two unknowns – the emissivity and the temperature, and the main difficulty is to calculate/estimate these two variables separately and reliably. Several methods have been developed in the recent decades to mitigate this problem. However, regardless of the method, all have developed similar incapacities for specific applications such as the type of sensor, study type, the target in question, the number of targets, type of weather, among others. Thus, the method to be applied depends on the study in question and the best results can be reached choosing the best fit method for that problem. In this work, we propose an alternative approach for estimating the temperature, and therefore the emissivity, of a particular target. The approach consists of generating statistical regressions in some wavelengths from linearized radiance function of laboratory data from a quartz sample at different temperatures, measured under controlled conditions of humidity and room temperature. The aim of regressions is to model the variation in temperature due to the variations in the radiance of the target in order to estimate the temperature from radiance data on a certain wavelength and without prior knowledge of the target emissivity. Laboratory datasets were divided into two groups - training and control. In the training group, several polynomial regressions were applied while the control group served to validate and evaluate the regressions. Five experiments were performed: (1) laboratory data at specific wavelengths (2) the central wave lengths of ASTER-TIR bands (3) simulations of ASTER-TIR bands (4) simulation of the atmosphere (dry and wet) for simulated bands of ASTER-TIR and (5) an image L1B ASTER-TIR of the study area validated with the AST08 product. As a result, it was possible to estimate the temperature with errors less than 0.2K from laboratory data and with mean error less than 1.5K from L1B ASTER-TIR images. Furthermore, the method requires only a spectral band in the image, enabling their application in monospectral thermal sensors. Satisfactory results were obtained with a simple linear regression and improved by increasing the wavelength. However, increasing the wavelength and, simultaneously, the degree of polynomial regression the results also improve with respect to linear regression results, but this improvement is insignificant, and thus the linear fit is the best option. Thus, the proposed method has shown promise, signaling that further research is needed.

Denna rapport presenterar ett alternativt sätt att ljussätta terräng i datorgrafik. Tidigare modeller har vanligtvis byggt på lokal ljussättning, som inte tar hänsyn till kringliggande geometri, och har med en extra process approximerat effekten av ljusinteraktionen. Genom att använda sig av en teknik som kallas precomputed radiance transfer (PRT) kan man förberäkna hur en punkt interagerar med ljus för olika inkommande riktningar och undviker därmed att göra detta under programkörningen. Det är viktigt att denna teknik även fungerar tillsammans med level-of-detail (LOD) terrängrenderingsalgoritmer eftersom rendering av alla trianglar i terrängen för varje skärmuppdatering inte är optimalt för dagens grafikkort. Man vill därför representera den underliggande terrängen med fler trianglar närmare betraktaren och färre längre bort. Motiveringen till detta är att trianglar längre ifrån betraktaren kommer resultera i färre pixlar på skärmen, så att rendera större trianglar gör en liten visuell skillnad. Arbetet visar på att tekniken fungerar med LOD-terrängrenderingsalgoritmer med bra prestanda och visuell kvalitet.

La simulation d'éclairage est un processus qui s'avère plus complexe (temps de calcul, coût mémoire, mise en œuvre complexe) aussi bien pour les matériaux brillants que pour les matériaux lambertiens ou spéculaires. Afin d'éviter le calcul coûteux de certains termes de l'équation de luminance (convolution entre la fonction de réflexion des matériaux et la distribution de luminance de l'environnement), nous proposons une nouvelle structure de données appelée Source Surfacique Équivalente (SSE). L'utilisation de cette structure de données nécessite le pré-calcul puis la modélisation du comportement des matériaux soumis à divers types de sources lumineuses (positions, étendues). L'exploitation d'algorithmes génétiques nous permet de déterminer les paramètres des modèles de BRDF, en introduisant une première source d'approximation. L'approche de simulation d'éclairage utilisée est basée sur un cache de luminance. Ce dernier consiste à stocker l'éclairement incident sous forme de SSE en des points appelés enregistrements. Durant la simulation d'éclairage, l'environnement lumineux doit également être assimilé à un ensemble de sources surfaciques équivalentes (en chaque enregistrement) qu'il convient de définir de manière dynamique. Cette phase constitue une deuxième source d'erreur. Toutefois, l'incertitude globale ne se réduit pas au cumul des approximations réalisées à chaque étape. Les comparatives réalisées prouvent, au contraire, que l'approche des Sources Surfaciques Équivalentes est particulièrement intéressante pour des matériaux rugueux ou pour les matériaux très brillants placés dans des environnements relativement uniformes. L'utilisation de SSE a permis de réduire considérablement à la fois le coût mémoire et le temps de calcul. Une fois que les SSE sont calculés en chaque enregistrement et pour un certain nombre de points de vue, nous proposons une nouvelle méthode de visualisation interactive exploitant les performances des GPU (carte graphique) et s'avérant plus rapide que les méthodes existantes. Enfin nous traiterons le cas où les grandeurs photométriques sont spectrales, ce qui est très important lorsqu'il s'agit de réaliser des simulations d'éclairage précises. Nous montrerons comment adapter les zones d'influence des enregistrements en fonction des gradients de luminance et de la géométrie autour des enregistrements. / Radiance caching methods have proven efficient for global illumination. Their goal is to compute precisely illumination values (incident radiance or irradiance) at a reasonable number of points lying on the scene surfaces. These points, called records, are stored in a cache used for estimating illumination of other points in the scene. Unfortunately, with records lying on glossy surfaces, the irradiance value alone is not sufficient to evaluate the reflected radiance; each record should also store the incident radiance for all incident directions. Memory storage can be reduced with projection techniques using spherical harmonics or other basis functions. These techniques provide good results with low shininess BRDFs. However, they get impractical for shininess of even moderate value since the number of projection coefficients increase drastically. In this paper, we propose a new radiance caching method, that handles highly glossy surfaces, while requiring a low memory storage. Each cache record stores a coarse representation of the incident illumination thanks to a new data structure called Equivalent Area light Sources (EAS), capable of handling fuzzy mirror surfaces. In addition, our method proposes a new simplification of the interpolation process since it avoids the need for expressing and evaluating complex gradients. Moreover, we propose a new GPU based visualisation method which exploits these EAS data structure. Thus, interactive rendering is done faster than existing methods. Finally, physical ligting simulations need to manipulate spectral physical quantities. We demonstrate in our work how these quantities can be handle with our technic by adapting the record influence zone depending on the radiance gradients and the geometry around the records.

Computing complex lighting simulations such as global illumination is a computationally intensive task. Various real time solutions exist to approximate aspects of global illumination such as shadows, however, few of these methods offer single pass rendering solutions for soft shadows (self and other) and inter-reflections. In contrast, Precomputed Radiance Transfer (PRT) is a real-time computer graphics technique which pre-calculates an object's response to potential incident light. At run time, the actual incident light can be used to quickly illuminate the surface, rendering effects such as soft self-shadows and inter-reflections. In this thesis, we show that by calculating PRT lighting coefficients densely over a surface as texture data, additional surface detail can be encoded by integrating other computer graphics techniques, such as normal mapping. By calculating transfer coefficients densely over the surface of a mesh as texture data, greater fidelity can be achieved in lighting coarse meshes than simple interpolation can achieve. Furthermore, the lighting on low polygon objects can be enhanced by drawing surface normal and occlusion data from highly tessellated, detailed meshes. By applying such data to a decimated, simplified mesh, a more detailed and visually pleasing reconstruction can be displayed for a lower cost. In addition, this thesis introduces Hierarchical PRT, which extends some surface effects, such as soft shadows, between objects. Previous approaches to PRT used a more complex neighborhood transfer scheme in order to extend these lighting effects. Hierarchical PRT attempts to capture scene information in a tree data structure which represents coarse lighting relationships between objects. Potential occlusions can be found at run time by utilizing the same spherical harmonic representation used to represent surface lighting to instead store light "filters" between scene tree nodes. Such "filters" can be combined over a set of nodes in the scene to obtain the net shadowing of an object with good performance. We present both visually pleasing results on simplified meshes using normal mapping and textured PRT and initial results using Hierarchical PRT that captures low frequency lighting information for a small number of dynamic objects which shadow static scene objects with good results.

Beginning with the discovery of graphene, two-dimensional materials have amassed a strong interest. Like graphene, transition metal dichalcogenides (TMDs) can be coaxed into atomically thin sheets which have some impressive properties. Unlike graphene, TMDs also has a change in their electronic band structure causing an indirect band gap to a direct gap transition in its monolayer form. Additionally, these materials lose their inversion symmetry as a monolayer. These unique properties make TMDs a strong candidate for being used in optoelectronic and valleytronic devices. In order for these devices to be successful, the optical properties of TMDs must be thoroughly understood. Due to this class of material's strong Coulomb interaction, the optical properties are dominated by excitons, a quasiparticle made up of an electron-hole pair. Therefore, the success of these devices relies, in part, on understanding and manipulating excitons. One key parameter of excitons is their dephasing rate which characterizes the lifetime of the coherent superposition of two states (i.e. how the coherence decays which is caused by excitons interacting with their environment). In this work, two optical properties are investigated: (1) How the linear absorption of the TMDs A-exciton peak varies as the material increases in thickness. By looking at how the absorption varies by sample thickness, the interaction between emitters can be understood. Experimental results for the diamagnetic shift are presented which are used to determine the lateral excitonic size. Through theoretical calculations, based on the semiconductor Maxwell-Bloch equations, additional insight into the radiative coupling of the systems are obtained. (2) How the coherence prole of the exciton changes in the presence of an external magnetic eld and specic valley excitation. By varying the polarization scheme in the four wave mixing measurement, specic valley excitation can be selected, allowing for insight into the dephasing mechanisms. By applying an external magnetic eld, the energy levels of the electron and hole can be discretized and the corresponding eects on the system's coherence seen. In conjunction with time-dependent density function theory calculations and the experimental results, a deeper understanding of exciton dynamics and multi-exciton complexes was obtained. Finally, a new system is proposed in which complex spectroscopic techniques can be performed on micron sized samples as well as devices in the presence of an external magnetic eld at cryogen temperatures. This system will allow for the investigation of the optical properties of stacked monolayers (heterostructures) as well as devices.

<p> </p> <p>As a result of Shadow NeRF and Sat-NeRF, it is possible to take the solar angle into account in a NeRF-based framework for rendering a scene from a novel viewpoint using satellite images for training. Our work extends those contributions and shows how one can make the renderings season-specific. Our main challenge was creating a Neural Radiance Field (NeRF) that could render seasonal features independently of viewing angle and solar angle</p> <p>while still being able to render shadows. We teach our network to render seasonal features by introducing one more input variable — time of the year. However, the small training datasets typical of satellite imagery can introduce ambiguities in cases where shadows are present in the same location for every image of a particular season. We add additional terms to the loss function to discourage the network from using seasonal features for accounting for shadows. We show the performance of our network on eight Areas of Interest containing images captured by the Maxar WorldView-3 satellite. This evaluation includes tests measuring the ability of our framework to accurately render novel views, generate height maps, predict shadows, and specify seasonal features independently from shadows. Our ablation</p> <p>studies justify the choices made for network design parameters. Also included in this work is a novel approach to space carving which merges multiple features and consistency metrics</p> <p>at different spatial scales to create higher quality digital surface map than is possible using standard RGB features.</p>

In modern society, cameras on intelligent devices can generate a huge amount of natural images, including images of the human body and face. Therefore, there is a huge social demand for more efficient editing of images to meet human production and life needs, including entertainment, such as image beauty. In recent years, Generative Models with Deep Learning techniques have attracted lots of attention in the Artificial Intelligence field, and some powerful methods, such as Variational Autoencoder and Generative Adversarial Networks, can generate very high-resolution and realistic images, especially for facial images, human body image. In this thesis, we follow the powerful generative model to achieve image generation and editing tasks, and we focus on human image generation and editing tasks, including local eye and face generation and editing, global human body generation, and editing. We introduce different methods to improve previous baselines based on different human regions. 1) Eye region of human image: Gaze correction and redirection aim to manipulate the eye gaze to a desired direction. Previous common gaze correction methods require annotating training data with precise gaze and head pose information. To address this issue, we proposed the new datasets as training data and formulated the gaze correction task as a generative inpainting problem, addressed using two new modules. 2) Face region of human image: Based on a powerful generative model for face region, many papers have learned to control the latent space to manipulate face attributes. However, they need more precise controls on 3d factors such as camera pose because they tend to ignore the underlying 3D scene rendering process. Thus, we take the pre-trained 3D-Aware generative model as the backbone and learn to manipulate the latent space using the attribute labels as conditional information to achieve the 3D-Aware face generation and editing task. 3) Human Body region of human image: 3D-Aware generative models have been shown to produce realistic images representing rigid/semi-rigid objects, such as facial regions. However, they usually struggle to generate high-quality images representing non-rigid objects, such as the human body, which greatly interests many computer graphics applications. Thus, we introduce semantic segmentation into the model. We split the entire generation pipeline into two stages and use intermediate segmentation masks to bridge these two stages. Furthermore, our model can control pose, semantic, and appearance codes by using multiple latent codes to achieve human image editing.

40 light years from Earth an Earth sized exoplanet called Trappist 1d orbits the M-dwarf star called Trappist 1. Trappist 1d is located in the habitable zone where liquid water could exist on the surface of the planet which raises the question: Could Trappist 1d be habitable? Since it is not known what the planet looks like, several simulations of potentially habitable climates were made including different water levels and atmospheric pressures with Earth-like atmospheres. Real observations with JWST and VLT are currently being made for the light passing through Trappist 1d’s potential atmosphere. In order to interpret the data and make any conclusions about the habitability of Trappist 1d, simulated spectra need to be made for the different scenarios. The goal of this project was to produce radiance spectrum of how observations viewed through different instruments would look like for the different planetary scenarios. The result of the project gave spectra that were quite similar, but differed specifically in the depths of the lines, meaning that in theory it could be possible to distinguish between the planetary scenarios via observations. In reality, because of uncertainties in the observations, it is probably not possible to distinguish between the different planetary models, but it might be enough to conclude if the planet has an Earth like atmosphere containing CO2 and H2O or not. / 40 ljusår bort från jorden kretsar en jordlik planet vid namn Trappist 1d runt en röd dvärgstjärna. Trappist 1d ligger i den så kallade beboeliga zonen där det är möjligt för flytande vatten att existera på planetens yta. Detta medför frågan: Finns det förutsättningar för liv på Trappist 1d? Eftersom det inte är känt hur det ser ut på planeten har flera potentiellt beboeliga klimat simulerats för olika vattennivåer och atmosfärstryck med en jordlik atmosfär. Olika instrument på teleskopen JWST och VLT samlar för tillfället in data för observationer genom Trappist 1d:s potentiella atmosfär. För att kunna tolka datan och dra slutsatser om förutsättningarna för liv på Trappist 1d behövs simulerade spektrum att jämföra med. Målet med det här projektet är att producera simulerade radians spektrum för hur observationer med de olika instrumenten hade sett ut för de olika scenarierna. Resultatet gav spektrum som främst skiljde sig i djupet av linjerna i graferna, vilket betyder att i teorin är det möjligt att skilja mellan de olika scenariona för en observation. På grund av osäkerheter i observationen, är det troligtvis inte möjligt i praktiken att se exakt vilket scenario det tillhör, men det kan vara tillräckligt för att säga ifall planeten har en jordlik atmosfär som innehåller vatten och koldioxid eller ej.