Abstract: Generative adversarial networks (GANs) have enabled photorealistic image synthesis and editing. However, due to the high computational cost of large-scale generators (e.g., StyleGAN2), it usually takes seconds to see the results of a single edit on edge devices, prohibiting interactive user experience. In this paper, we take inspirations from modern rendering software and propose Anycost GAN for interactive natural image editing. We train the Anycost GAN to support elastic resolutions and channels for faster image generation at versatile speeds. Running subsets of the full generator produce outputs that are perceptually similar to the full generator, making them a good proxy for preview. By using sampling-based multi-resolution training, adaptive-channel training, and a generator-conditioned discriminator, the anycost generator can be evaluated at various configurations while achieving better image quality compared to separately trained models. Furthermore, we develop new encoder training and latent code optimization techniques to encourage consistency between the different sub-generators during image projection. Anycost GAN can be executed at various cost budgets (up to 10x computation reduction) and adapt to a wide range of hardware and latency requirements. When deployed on desktop CPUs and edge devices, our model can provide perceptually similar previews at 6-12x speedup, enabling interactive image editing. The code and demo are publicly available: this https URL.

Abstract: In this paper we introduce PyXYZ, a 3D wireframe software rendering framework for educational purposes. The main goal of this framework is to provide a simple-to-understand tool that students can use to build a more sophisticated engine, while learning mathematics and acquiring a deeper knowledge of the complexity of a modern 3D engine. PyXYZ can be used as a teaching aid in course work and/or as a template for multi-goal project assignments, allowing students with diverse capabilities and interests to have different levels of commitment. The engine has been used with positive results in a mathematics course unit of a computer games BA and can be easily adapted to various teaching scenarios.

Abstract: In this paper VisualEnv, a new tool for creating visual environment for reinforcement learning is introduced. It is the product of an integration of an open-source modelling and rendering software, Blender, and a python module used to generate environment model for simulation, OpenAI Gym. VisualEnv allows the user to create custom environments with photorealistic rendering capabilities and full integration with python. The framework is described and tested on a series of example problems that showcase its features for training reinforcement learning agents.

Abstract: It can be argued that software can be seen as a form of art and digital heritage and yet it rarely enjoys the same efforts afforded to it compared to physical counterparts. There are many reasons for this, such as the increasing costs of maintenance or the reducing amount of expertise in the specific aging technology. Maintaining software and ensuring that it continues to work on current hardware and operating systems is known as digital preservation. There are many ways in which we can attempt to preserve digital software and one of the most effective ones is by using emulation to simulate the obsolete hardware. However, for games and other entertainment media, this technique is not always effective due to a requirement on specific hardware, such as an accelerated GPU in order to reach an acceptable performance for the user. It is often difficult to emulate a GPU and, as such, a different approach often needs to be taken, which reduces the flexibility and portability of the emulation software. Hydra is a new approach to accessing the native hardware from within an emulated environment which allows for a much simpler emulator to be developed and maintained and yet also offers the potential of accessing other types of hardware without needing to modify the emulation software itself. Hydra is designed to be platform agnostic in that not only is it possible to integrate with existing emulators but also be immediately usable from within guest operating systems, ranging from legacy platforms such as MS-DOS, through to modern platforms such as the PlayStation 4 (Orbis OS, a FreeBSD derivative), through to more exotic platforms such as Plan 9 from Bell Laboratories. It can do this because it does not rely on a complex emulator-specific virtual driver stack. This PhD thesis provides the research undertaken for Hydra, including the motivation behind it, the specific problems it was designed to solve and how it can be implemented in a platform agnostic manner. Hydra’s performance is analysed to ascertain the suitability of the output to cater for, specifically, a wide variety of platforms that it can run on in a satisfactory manner within less powerful or emulated environments. A performance analysis study is conducted to ensure that the technology provides an acceptable solution to accessing preserved titles. This study concluded with results showing that Hydra offers a greater performance than software rendering, especially within emulated environments. A bandwidth comparison between Hydra and VNC was undertaken to ascertain the use of the technology as a streaming medium. The results concluded that under specific conditions, Hydra performed better than VNC by streaming at a higher resolution and consuming less bandwidth. Hydra is also utilised in a number of engineering tasks relating to preservation of software. The experiences of using Hydra in this way are discussed, including any difficulties encountered. Lastly, a conclusion is made and any future work is identified.

Abstract: A content server system is provided that includes at least one processor configured to store a user account for a user at a content server of the content server system that is configured to serve server-side rendered content to a client computing device of the user. Audio is played out via a sound output device associated with the client computing device. The at least one processor is further configured to determine a licensing identifier associated with a device of the user or the user account of the user, send the licensing identifier to a third-party licensing server device, receive an indication that an active license is associated with the licensing identifier, determine that the user is authorized to access a digital rights managed audio rendering software, and cause audio of the server-side rendered content to be rendered using audio rendering algorithms of the digital rights managed audio rendering software.

Abstract: Generative adversarial networks (GANs) have enabled photorealistic image synthesis and editing. However, due to the high computational cost of large-scale generators (e.g., StyleGAN2), it usually takes seconds to see the results of a single edit on edge devices, prohibiting interactive user experience. In this paper, inspired by quick preview features in modern rendering software, we propose Anycost GAN for interactive natural image editing. We train the Anycost GAN to support elastic resolutions and channels for faster image generation at versatile speeds. Running subsets of the full generator produce outputs that are perceptually similar to the full generator, making them a good proxy for quick preview. By using sampling-based multi-resolution training, adaptive-channel training, and a generator-conditioned discriminator, the anycost generator can be evaluated at various configurations while achieving better image quality compared to separately trained models. Furthermore, we develop new encoder training and latent code optimization techniques to encourage consistency between the different sub-generators during image projection. Anycost GAN can be executed at various cost budgets (up to 10× computation reduction) and adapt to a wide range of hardware and la tency requirements. When deployed on desktop CPUs and edge devices, our model can provide perceptually similar previews at 6-12× speedup, enabling interactive image editing. The ${\color{RubineRed}{code}}$ and ${\color{RubineRed}{demo}}$ are publicly available.

Abstract: <p><span>Blazor, a web framework developed by Microsoft, offers two distinct approaches for rendering web pages: Server-Side Rendering (SSR) and Client-Side Rendering (CSR). This paper aims to evaluate the performance, scalability, and user experience of these two rendering techniques in Blazor. Through a series of experiments and performance benchmarks, we analyze the strengths and weaknesses of SSR and CSR, providing insights for developers on choosing the appropriate rendering strategy for their applications.</span></p>