Browsing by Author "Bemana, Mojtaba"

Now showing 1 - 3 of 3
  • No Thumbnail Available
    Item
    Efficient image-based rendering
    (2023-07-12) Bemana, Mojtaba
    Recent advancements in real-time ray tracing and deep learning have significantly enhanced the realism of computer-generated images. However, conventional 3D computer graphics (CG) can still be time-consuming and resource-intensive, particularly when creating photo-realistic simulations of complex or animated scenes. Image-based rendering (IBR) has emerged as an alternative approach that utilizes pre-captured images from the real world to generate realistic images in real-time, eliminating the need for extensive modeling. Although IBR has its advantages, it faces challenges in providing the same level of control over scene attributes as traditional CG pipelines and accurately reproducing complex scenes and objects with different materials, such as transparent objects. This thesis endeavors to address these issues by harnessing the power of deep learning and incorporating the fundamental principles of graphics and physical-based rendering. It offers an efficient solution that enables interactive manipulation of real-world dynamic scenes captured from sparse views, lighting positions, and times, as well as a physically-based approach that facilitates accurate reproduction of the view dependency effect resulting from the interaction between transparent objects and their surrounding environment. Additionally, this thesis develops a visibility metric that can identify artifacts in the reconstructed IBR images without observing the reference image, thereby contributing to the design of an effective IBR acquisition pipeline. Lastly, a perception-driven rendering technique is developed to provide high-fidelity visual content in virtual reality displays while retaining computational efficiency.
  • Thumbnail Image
    Item
    Enhancing Image Quality Prediction with Self-supervised Visual Masking
    (The Eurographics Association and John Wiley & Sons Ltd., 2024) Çogalan, Ugur; Bemana, Mojtaba; Seidel, Hans-Peter; Myszkowski, Karol; Bermano, Amit H.; Kalogerakis, Evangelos
    Full-reference image quality metrics (FR-IQMs) aim to measure the visual differences between a pair of reference and distorted images, with the goal of accurately predicting human judgments. However, existing FR-IQMs, including traditional ones like PSNR and SSIM and even perceptual ones such as HDR-VDP, LPIPS, and DISTS, still fall short in capturing the complexities and nuances of human perception. In this work, rather than devising a novel IQM model, we seek to improve upon the perceptual quality of existing FR-IQM methods. We achieve this by considering visual masking, an important characteristic of the human visual system that changes its sensitivity to distortions as a function of local image content. Specifically, for a given FR-IQM metric, we propose to predict a visual masking model that modulates reference and distorted images in a way that penalizes the visual errors based on their visibility. Since the ground truth visual masks are difficult to obtain, we demonstrate how they can be derived in a self-supervised manner solely based on mean opinion scores (MOS) collected from an FR-IQM dataset. Our approach results in enhanced FR-IQM metrics that are more in line with human prediction both visually and quantitatively.
  • Thumbnail Image
    Item
    Video Frame Interpolation for High Dynamic Range Sequences Captured with Dual-exposure Sensors
    (The Eurographics Association and John Wiley & Sons Ltd., 2023) Cogalan, Ugur; Bemana, Mojtaba; Seidel, Hans-Peter; Myszkowski, Karol; Myszkowski, Karol; Niessner, Matthias
    Video frame interpolation (VFI) enables many important applications such as slow motion playback and frame rate conversion. However, one major challenge in using VFI is accurately handling high dynamic range (HDR) scenes with complex motion. To this end, we explore the possible advantages of dual-exposure sensors that readily provide sharp short and blurry long exposures that are spatially registered and whose ends are temporally aligned. This way, motion blur registers temporally continuous information on the scene motion that, combined with the sharp reference, enables more precise motion sampling within a single camera shot. We demonstrate that this facilitates a more complex motion reconstruction in the VFI task, as well as HDR frame reconstruction that so far has been considered only for the originally captured frames, not in-between interpolated frames. We design a neural network trained in these tasks that clearly outperforms existing solutions. We also propose a metric for scene motion complexity that provides important insights into the performance of VFI methods at test time.